MIGRATION, DISPERSAL, AND SURVIVAL PATTERNS OF MULE DEER

(ODOCOILEUS HEMIONUS) IN A CHRONIC WASTING DISEASE-ENDEMIC AREA OF

SOUTHERN SASKATCHEWAN

A Thesis Submitted to the College of

Graduate Studies and Research

In Partial Fulfillment of the Requirements

For the Degree of Master of Science

In the Department of Veterinary Pathology

University of Saskatchewan

Saskatoon

By

NICOLE K. SKELTON

Copyright Nicole K. Skelton, September, 2010. All rights reserved.

i

PERMISSION TO USE

In presenting this thesis in partial fulfilment of the requirements for a Postgraduate degree from

the University of Saskatchewan, I agree that the Libraries of this University may make it freely

available for inspection. I further agree that permission for copying of this thesis in any manner,

in whole or in part, for scholarly purposes may be granted by the professor or professors who

supervised my thesis work or, in their absence, by the Head of the Department or the Dean of the

College in which my thesis work was done. It is understood that any copying or publication or

use of this thesis or parts thereof for financial gain shall not be allowed without my written

permission. It is also understood that due recognition shall be given to me and to the University

of Saskatchewan in any scholarly use which may be made of any material in my thesis.

Requests for permission to copy or to make other use of material in this thesis in

whole or part should be addressed to:

Head of the Department of Veterinary Pathology

Western College of Veterinary Medicine

University of Saskatchewan

52 Campus Drive

Saskatoon, Saskatchewan S7N 5B4

ii

ABSTRACT

Chronic wasting disease (CWD) has infected wild deer of Saskatchewan for at least the

past 10 years. Disease management plans have evolved over the years, but without information

on mule deer (Odocoileus hemionus) habits and movements in the grasslands of southern

Saskatchewan. We radio-collared and monitored the survival and movements of 206 mule deer

from 2006 to 2009. Long distance movements by deer have potential to transfer disease to

previously naïve areas. Survival rates had not yet been evaluated in this area; baseline data will

provide a useful measure for population-level impacts of the disease in the future.

Juvenile dispersals and adult migrations were contrasted from 4 study areas along the

South Saskatchewan River. Dispersal distance (median = 22.8 km, n = 14) was similar to

migration distance (median = 16.0 km, n = 49). Median migration distance was similar between

males (15.7 km, n = 51) and females (19.7 km, n = 65). Obligatory migrants were more likely to

be female. Deer from an area of extensive grassland were more likely to be migratory than their

counterparts in fragmented grassland of an agricultural landscape. Maximum migration and

dispersal distances were 113 km and 195 km, respectively. Movement paths of 33 GPS-collared

deer were best explained by high terrain ruggedness values and proximity to grassland.

Seasonal survival rates showed that deer had lowest survival in autumn months during

hunting season. Juveniles had lower survival than adults in all seasons. Harvest regime changes

in 2008 improved the autumn survival of adult females but adult males had lower survival than

in 2007. Body condition of captured deer was evaluated from residuals of mass-length

regression. Cox regression analyses suggested that deer in good body condition (75th percentile)

were half as likely to die and that those in very poor body condition (10th percentile) were twice

iii

as likely to die. Radio collars that weighed > 2% of body mass negatively affected survival and

we recommend future researchers take this into consideration.

Survival, dispersal, and migration rates and patterns are crucial parameters in modeling

CWD in local mule deer populations. Saskatchewan wildlife managers aim to prevent CWD

spread into new areas, and can use mule deer movement orientations to target surveillance

accordingly. White-tailed deer (Odocoileus virginianus) movements are briefly discussed;

further knowledge of their movements is required for CWD management in all of Saskatchewan.

iv

ACKNOWLEDGEMENTS

An extensive list of people was involved in this research project. I am grateful to my supervisor,

Trent Bollinger, for welcoming me to become a part of it. His knowledge and guidance were well

balanced with an ability to let me develop my own skills, and his support and encouragement allowed me

to endeavor when my confidence waned. Thank you to my advisory committee: Drs. Tasha Epp and Phil

McLoughlin for their assistance, Dr. Ray Alisauskas for accepting the role of external examiner, and

graduate chairs Drs. Marion Jackson, Gary Wobeser, and Beverly Kidney. Dr. Cheryl Waldner provided

help with statistical analyses. Dr. Marc Cattet (CCWHC) shared his experience with body condition

indices and Dr. Xulin Guo welcomed all of my remote sensing inquiries.

Erin Silbernagel was my partner in all aspects of this research, and I thank her for all of her ideas,

efforts, and determination—and more than that she was a great friend. CCWHC employees Marnie

Zimmer and Christine Wilson provided endless support in the field and in the office, and no doubt saved

us all from a few minor emergencies. I thank the capture crew, V. Harris, T. Quirk, and D. Harder for

their hard work and tolerating my requests—even if I handed them a sewing needle and thread after

supper. J. Meaden, A. Curtis, and J. Skelton contributed to radiotelemetry work and collar removal at the

end of the research. Fellow graduate students in Vet Pathology and Biology could be relied on for lively

discussion and camaraderie. Within the WCVM I thank I. Shirley, S. Mayes, N. Kozakevich, J.

Diedrichs, and T. Moss for all their administrative and logistical support.

Finally, thank you to my friends and family for your enduring love and support. My mom Jacqui

supported me always, and offered her home and a warm meal on late nights. To my husband Jeremi,

thank you for your encouragement. Your passion for all things deer-related inspired me to take on this

project. I only hope I can support your ambitions with as much patience and enthusiasm as you have

mine.

This research was made possible by funding and support from PrioNet Canada, Saskatchewan

Ministry of Environment, and Canadian Cooperative Wildlife Health Centre. We are grateful for the

cooperation of residents and landowners of the communities of Beechy, Elbow, Kyle, Matador Farming

Pool, Stewart Valley, and Cabri. Mitchinsen Flying Service Ltd. provided flights for radio-tracking and

both Bighorn Helicopter Inc., and QuickSilver Air Inc. provided netgunning crews. Dennis Duro

completed the landcover map.

This thesis is dedicated to my late father, Robert Koskie. Everything I have achieved in my life is because you led by example to never limit oneself; and that life’s opportunities are earned through perseverance, hard work, and the will and desire to follow one’s passion and instinct. You instilled in me your awe of nature, humility, strength, and the foundation to become the person I am today. You are missed but ever-present.

vi

TABLE OF CONTENTS

PERMISSION TO USE .............................................................................................................. i ABSTRACT .............................................................................................................................. ii ACKNOWLEDGEMENTS........................................................................................................iv TABLE OF CONTENTS ...........................................................................................................vi LIST OF TABLES .....................................................................................................................ix LIST OF FIGURES ....................................................................................................................x LIST OF ABBREVIATIONS.....................................................................................................xi CHAPTER 1 INTRODUCTION AND LITERATURE REVIEW .........................................1 1.1 Chronic Wasting Disease.......................................................................................................1

1.1.1 CWD in Saskatchewan ...................................................................................................1 1.1.2 CWD Management Challenges .......................................................................................3

1.2 Literature Review..................................................................................................................4 1.2.1 Study Species: Mule Deer (Odocoileus hemionus) ..........................................................4

1.2.1.1 Habitat preferences ..................................................................................................5 1.2.1.2 Breeding & social structure......................................................................................6 1.2.1.3 Importance to society...............................................................................................6

1.2.2 Dispersal ........................................................................................................................7 1.2.3 Migration........................................................................................................................8 1.2.4 Survival ..........................................................................................................................9

1.2.4.1 Body condition ......................................................................................................10 1.2.4.2 Home range habitat and survival ............................................................................11 1.2.4.3 Radio collar effects on survival ..............................................................................11

1.3 Objectives ...........................................................................................................................12 CHAPTER 2 DISPERSAL AND MIGRATION OF WILD DEER IN A CWD-ENDEMIC AREA OF SOUTHERN SASKATCHEWAN........................................................................13 2.1 Abstract...............................................................................................................................13 2.2 Introduction.........................................................................................................................14 2.3 Methods ..............................................................................................................................16

2.3.1 Study Area....................................................................................................................16 2.3.2 Capture.........................................................................................................................21 2.3.3 Definition and Measurement of Long Distance Movements ..........................................22 2.3.4 Dispersal Rate ..............................................................................................................24 2.3.5 Factors Associated with Long Distance Movements .....................................................24 2.3.6 Statistical Analyses.......................................................................................................25

2.4 Results ................................................................................................................................28

vii

2.4.1 Long Distance Movements ...........................................................................................28 2.4.1.1 Dispersal................................................................................................................28 2.4.1.2 Migration...............................................................................................................32 2.4.1.3 Dispersal vs. migration ..........................................................................................34 2.4.1.4 Excursions .............................................................................................................35

2.4.2 Attributes of Long Distance Movements.......................................................................39 2.4.2.1 Travel speed ..........................................................................................................39 2.4.2.2 Travel orientation...................................................................................................41 2.4.2.3 Habitat selection ....................................................................................................45

2.5 Discussion...........................................................................................................................45 2.5.1 Long Distance Movement Demographics .....................................................................45 2.5.2 Habitat and Long Distance Movements.........................................................................46

2.6 Management Implications: CWD and Long Distance Movements .......................................48 2.6.1 Mule Deer ....................................................................................................................48 2.6.2 White-tailed Deer and CWD in Saskatchewan ..............................................................50

CHAPTER 3 MULE DEER (ODOCOILEUS HEMIONUS) SURVIVAL RATES IN A CWD-ENDEMIC AREA OF SOUTHERN SASKATCHEWAN .........................................52 3.1 Abstract...............................................................................................................................52 3.2 Introduction.........................................................................................................................53 3.3 Methods ..............................................................................................................................54

3.3.1 Study Area....................................................................................................................54 3.3.2 Capture.........................................................................................................................54 3.3.3 Radio Collars................................................................................................................54 3.3.4 Body Condition Index (BCI).........................................................................................54 3.3.5 Home Range Attributes ................................................................................................55

3.3.5.1 Land cover classification map ................................................................................55 3.3.5.2 Home range calculation..........................................................................................56

3.3.6 Statistical Analyses.......................................................................................................57 3.3.6.1 Dataset...................................................................................................................57 3.3.6.2 Kaplan-Meier estimates .........................................................................................57 3.3.6.3 Cox regression .......................................................................................................58 3.3.6.4 Post-hoc winter severity analysis............................................................................61

3.4 Results ................................................................................................................................61 3.4.1 Capture.........................................................................................................................61 3.4.2 CWD Results................................................................................................................62 3.4.3 Body Measures .............................................................................................................62 3.4.4 Body Condition Index...................................................................................................63 3.4.5 Radio Collars................................................................................................................65 3.4.5 Survival Estimates ........................................................................................................66

3.4.5.1 Kaplan-Meier estimates .........................................................................................66 3.4.5.1.1 Adults .................................................................................................................66

Annual...........................................................................................................................66 Seasonal ........................................................................................................................66

3.4.5.1.2 Juveniles .............................................................................................................68 Annual...........................................................................................................................68

viii

Seasonal ........................................................................................................................68 3.4.6.1 Cox regression...........................................................................................................68

3.4.6.1.1 Full dataset..........................................................................................................68 3.4.6.3.1.2 Body condition subset ......................................................................................69 3.4.6.3.1.3 Home range subset ...........................................................................................70 3.4.6.3.1.4 Winter severity.................................................................................................70

3.5 Discussion...........................................................................................................................71 3.5.1 CWD Management Program.........................................................................................71

3.5.1.1 Male and female survival during the hunting season ..............................................71 3.5.2 Body Condition ............................................................................................................73 3.5.3 Radio Collars................................................................................................................74 3.5.4 Home Range Habitat Effect on Survival .......................................................................74

3.6 Management Implications ...................................................................................................75 CHAPTER 4 SYNTHESIS ....................................................................................................77 4.1 Study Limitations ................................................................................................................79 4.2 Conclusion ..........................................................................................................................80 Literature cited..........................................................................................................................83 Appendix ..................................................................................................................................93

A. Accuracy Assessment of Land Cover Map.....................................................................93

ix

LIST OF TABLES

Table page 2.1 Dispersal details .......................................................................................................... 29

2.2 Juveniles and dispersal by study site ............................................................................ 30

2.3 Number of migratory and resident deer by study site and sex....................................... 33

2.4 Migration distances by study site and sex .................................................................... 33

2.5 Excursion distance by site ........................................................................................... 35

2.6 Excursion distance by month....................................................................................... 35

2.7 Period of excursion by sex........................................................................................... 37

2.8 Travel speed and duration of dispersal for GPS-collared mule deer ............................. 39

2.9 Travel speed and duration of migration for GPS-collared mule deer ............................ 40

2.10 Mean migration vector by study site .......................................................................... 41

2.11 GEE model results..................................................................................................... 45

3.1 Body condition categories based on percentiles ........................................................... 59

3.2 Body measures of captured mule deer ......................................................................... 62

3.3 Factor loadings of morphologic measures on PC1 ....................................................... 63

3.4 Annual (Apr-Mar) survival rates of adult male and female radio-collared mule deer.... 66

3.5 Adult mule deer seasonal survival 2006-2008.............................................................. 67

3.6 Survival of juvenile mule deer captured in 2007 .......................................................... 68

3.7 Model results of Cox regression analysis .................................................................... 70

3.8 Cox regression model parameter estimates and hazard ratios ...................................... 70

3.9 Comparison of winter severity indicators for 2007 and 2008 and 30-year normals ...... 71

x

LIST OF FIGURES

Figure page 2.1 Map of study sites in Southern Saskatchewan along the South Saskatchewan River .... 18

2.2 Total landcover area for each study site ....................................................................... 19

2.3 Proportional landcover for each study site ................................................................... 19

2.4 Locations of CWD positive wild deer in Saskatchewan ............................................... 20

2.5 Inset of previous map showing study area.................................................................... 21

2.6 Frequency histograms of dispersal distance and migration distance ............................. 26

2.7 Annual juvenile dispersal rate for 2007 and 2008 cohorts ............................................ 31

2.8 Overall juvenile dispersal rate 2007-2008.................................................................... 31

2.9 Juvenile dispersals from the ANT site......................................................................... 32

2.10 Mean excursion distance per month........................................................................... 36

2.11 Count of excursions made by mule deer for each month ............................................ 37

2.12 Period of mule deer excursions by sex ...................................................................... 38

2.13 Distance and direction (km) of dispersal events ......................................................... 42

2.14 Distance and direction (km) of migration events ....................................................... 43

2.15 Distance and direction (km) of excursion events........................................................ 44

3.1 Relationship between body size and body mass ........................................................... 64

3.2 Body condition index boxplot..................................................................................... 65

3.3 Seasonal survival rates of adult female and male mule deer ........................................ 67

xi

LIST OF ABBREVIATIONS

AD Adult

AIC Akaike information criterion

ANT Antelope Creek study site

BEEMAT Beechy and Matador pasture study site

BCI Body condition index

CCWHC Canadian Cooperative Wildlife Health Centre

CWD Chronic wasting disease

DOU Douglas Park study site

SWI Swift Current Creek study site

F Female

GEE Generalized estimating equation

GIS Geographic information system

GPS Global positioning system

JUV Juvenile

KDE Kernel density estimate

M Male

MOE Ministry of Environment

QIC Quasi-likelihood under independence model criterion

QICC Corrected quasi-likelihood under independence criterion

S.S. RIVER South Saskatchewan River

TRI Terrain ruggedness index

VHF Very high frequency

1

CHAPTER 1 INTRODUCTION AND LITERATURE REVIEW

1.1 Chronic Wasting Disease

Chronic wasting disease (CWD) is a fatal transmissible spongiform encephalopathy

(TSE) that affects wild and domestic cervids of North America. Like other prion diseases, it is

characterized by chronic neurodegeneration, loss of motor skills, and certain fatality as a result of

a protease-resistant isoform of cellular prions (PrPres) that accumulate in the host central nervous

system (CNS) and lymphatic tissues. Unlike most prion diseases (except scrapie), CWD can be

transmitted by environmental contamination (Mathiason et al. 2009) and like scrapie is

transmitted horizontally. Prions are found in bodily fluids (saliva, blood, urine) and excrement

(Mathiason et al. 2006), muscle tissue, and carcasses (Miller et al. 2004). Prions are notably

resistant to degradation in the environment (Johnson et al. 2006, Wiggins 2009). Disease spread

is relatively slow but persistent (Bollinger et al. 2004); once established in an area, eradication of

CWD has so far proven to be impossible. North American wildlife managers have struggled to

understand, manage, and geographically contain the disease for over 4 decades. It was first

documented in captive mule deer at a research facility in Colorado in 1967 and in free-ranging

elk of the same state in 1981 (Williams and Miller 2002). By early 2010, CWD was found in 17

states and 2 provinces.

1.1.1 CWD in Saskatchewan

A mule deer (Odocoileus hemionus) shot in 2000 in western Saskatchewan was the first

Canadian wild cervid to be detected with CWD. White-tailed deer (Odocoileus virginianus) now

2

account for a growing portion of CWD-positives in provincial surveillance and the disease was

recently detected in wild elk (Canadian Cooperative Wildlife Health Center (CCWHC),

unpublished data). Other cervid species of Canada susceptible to the disease include black-tailed

deer, moose (Alces alces americanus) and potentially caribou (Rangifer tarandus) (Happ et al.

2007).

Wildlife management is improved by scientific knowledge of local populations, but

disease outbreaks often require timely decisions by wildlife managers, often with incomplete

knowledge of important factors related to the ecology and epidemiology of the disease (Schauber

and Woolf 2003). Geographic spread of CWD can be predicted by host animal movement

(Miller et al. 2000). Mule deer life habits, including movement patterns, migration, dispersal, and

survival rates, vary by region and there are currently no publications on these topics in the

Canadian Prairies. The need for site-specific research in Saskatchewan arose in 2000 when

CWD was first detected in wild mule deer.

At that time, managers decided to reduce deer density where the positive was found and

to sample adjacent areas—an action plan similar to those already in place in areas of Wisconsin

and Colorado. The herd reduction program relied on hunter participation in the Earn-a-Buck

program: hunters were required to submit 2 antlerless (doe or fawn) heads in order to receive an

either-sex tag. The either-sex tag was usually used to harvest an adult male. Management efforts

have failed to eradicate the disease and in 2008 the Ministry of Environment made a decision to

shift focus from eradication to monitoring and reducing prevalence and minimizing spread

throughout the province. CWD was now considered enzootic in some areas.

At the time of this research project’s proposal, there was speculation that the Earn-a-Buck

management program, by encouraging harvest of females and adult males, was skewing local

3

population age/gender ratios toward higher proportions of juvenile males. Since juvenile males

are reportedly more likely to disperse, it was argued the program was likely increasing

geographic spread of disease. In response to these criticisms alterations were made to Earn-a-

Buck regulations allowing harvest of 3-point-or-less males, as well as females. This change

likely reduced the proportion of juvenile males in CWD control areas; nevertheless, the role of

dispersal in disease spread remained worthy of investigation. Young male deer are the highest

risk group for CWD infection (Osnas et al. 2009) and are also most likely to disperse

(Kammermeyer and Marchinton 1976, Dobson 1982, Nixon et al. 1994). These facts, in addition

to their often high numbers within a managed deer population, make juvenile males arguably of

utmost concern for geographic spread of CWD. During the pilot year of this study (2006), long

distance migrations were commonly observed at one study area. The role of migratory deer in

the geographic spread of disease across the landscape came into consideration.

1.1.2 CWD Management Challenges

CWD management programs for wild deer have developed rapidly over the past 20+

years in North America but clear evidence of efficacy has been difficult to demonstrate. The

programs mainly involve some form of deer cull through increased hunter harvest or sharp-

shooting. At the 3rd International CWD Symposium in 2009, a common concern expressed by

state and provincial wildlife managers was the decline of public support and diminished funds

for CWD management programs. Wildlife disease management programs are long-term

investments, and due to the slow spread and long time-course of CWD, detecting changes in

prevalence and distribution is difficult (Conner et al. 2008). Valid scientific assessment of the

efficacy of reducing deer density on CWD prevalence requires a large sample size and many

years of data (Conner et al. 2007). CWD management programs involving herd reduction are

4

controversial and are further complicated by the uncertain implications of CWD for wildlife,

domestic animal and human health (Williams and Miller 2002, Vaske 2010).

While funding and support may be declining, research activities over the past decade

have contributed a great deal of knowledge about CWD and managers have more tools to

understand and predict its transmission. With this knowledge the scientific community, wildlife

managers, and hunters can collaborate to innovatively manage deer and CWD. Advanced spatio-

temporal analyses, disease modeling, and lessons learned from the past will aid in efficient

allocation of funding to disease management programs. For example, Illinois has found a

significant decreasing trend of CWD prevalence in young deer where sharpshooting was

implemented for a number of years (Shelton et al. 2009).

Despite the challenges, CWD management programs continue because long-term effects

of the disease on wild populations are still unknown. In discrete areas, prevalence can be quite

high; it has been documented at >30% (Miller et al. 2008) in Colorado mule deer. Realizing

eradication is no longer a reasonable goal, many wildlife agencies (including Saskatchewan’s

Ministry of Environment) have shifted focus toward disease prevention in new areas, and

monitoring areas where it is currently found.

1.2 Literature Review

1.2.1 Study Species: Mule Deer (Odocoileus hemionus)

The mule deer is a North American cervid found in the western half of the continent,

extending in the south from Mexico and north into the Yukon of Canada. The species evolved in

the grassland and rugged badlands of the prairies and foothills of the Rocky and Sierra Mountain

ranges. Also known as black-tailed deer, there are 7 subspecies of O. hemionus recognized and 4

more debated (Mackie et al. 2003). The most widespread subspecies, the Rocky Mountain mule

5

deer (O. hemionus hemionus), is found in the mid-western United States and Canadian provinces

of Alberta, British Columbia, and Saskatchewan.

Mule deer share much of their range with their North American counterpart—the white-

tailed deer—that evolved in the deciduous forest areas of the east. They are easily distinguished

by their appearance and behavioral traits. Distinguishable characteristics include size, color and

shape of the tail, antler form, ear size, metatarsal gland size, and the mule deer’s distinctive

stotting gait—a 4-footed bound used to quickly navigate rugged terrain. They stott when

threatened, using precise movements to place themselves somewhere in their surroundings that

will provide cover or an obstacle to evade predators. White-tailed deer flee danger by quickly

running for the nearest cover, and so proximity to woodland cover is a more essential habitat

requirement than for mule deer.

1.2.1.1 Habitat preferences

Mule deer range extends across a number of ecoregions including semi-arid desert,

prairie, and mountain foothills; thus, they are adapted to a number of habitat types. Most often

they are associated with rugged terrain including steep slopes of mountainous areas, badlands,

and river breaks. The terrain may be shrub-covered, semi-forested, or open grassland.

Mule deer feed on herbaceous materials including leafy forbs, shrubs, and grasses, as

well as some browse (woody materials). Kufeld et al. (1973) listed 788 species of plants eaten by

Rocky Mountain mule deer (in Mackie et al. 2003). Mule deer acquire water from food sources

as well as free-water sources. They generally stay within a few kilometers of open water, and

females are more likely to be near water. In drought periods, fawn production is lower

(Lawrence et al. 2004).

6

1.2.1.2 Breeding & social structure

Deer are organized in matrilineal groups, with 2 or more generations of females in family

groups. The dominant or matriarch deer is a mature dam with successful reproductive history

(Hawkins and Klimstra 1970, Kucera 1978, Porter et al. 1991). Social organization varies

seasonally. Sexual segregation is common for all seasons except winter, when both sexes are

found in large groups on wintering range. During parturition, does isolate themselves to rear their

fawns (Ozaga et al. 1982). In the fall, they may reunite with previous offspring and other female

groups. Males tend to range in different areas than females and form buck groups (Mackie et al.

2003). Females exhibit more fidelity to their home ranges than do bucks (Geist 1994b).

Mule deer are polygynous breeders, but males tend to an individual female for short

periods until she is bred. Most begin breeding after the age of 1.5 years, and large bucks are

dominant breeders. The breeding period varies by location but is generally in autumn and early

winter; in Saskatchewan, breeding usually peaks in mid-November.

1.2.1.3 Importance to society

Mule deer provide aesthetic, recreational, and economic benefits to society and are

important species in ecosystem health. As large and abundant herbivores, they affect vegetation

composition and structure and are important to nutrient cycles (Augustine and McNaughton

1998). Deer are valued for recreation that is consumptive (i.e., hunting) as well as non-

consumptive (e.g., sight-seeing or photography). Recreational users are willing to pay to hunt or

just see deer (Conover 1997). They can also have economic cost through crop depredation,

landscape damage, and vehicle collisions (Kie and Czech 2000, Côté et al. 2004)

In Saskatchewan, big game species (including mule deer, white-tailed deer, elk, moose,

and black bears) are the primary choice of hunters. An economic evaluation of hunting in

Saskatchewan estimated annual expenditure by big-game hunters at over $30 million (Derek

7

Murray Consulting Associates 2006). Known presence of CWD in local deer is likely to alter

hunter behavior or participation (Miller and Shelby 2009), in turn affecting local and wildlife

agency revenue generated by hunters (Needham and Vaske 2008).

1.2.2 Dispersal

Dispersal is a means of genetic exchange essential to a species’ fitness. Local population

dynamics are a function of additions (births and immigrations) and losses (deaths and

emigrations). Despite its essential role in populations, dispersal is relatively poorly understood.

Much of the literature on North American deer movements refers to studies on white-tailed deer

in the United States. Deer dispersal is defined as a permanent movement away from the

individual’s natal range, and is usually undertaken by yearling males. Dispersal reduces

inbreeding and resource competition (Kammermeyer and Marchinton 1976, Holzenbein and

Marchinton 1992, Wolff 1993, Rosenberry et al. 2001).

Previous mule deer studies report high fidelity to annual or seasonal home ranges

(Garrott et al. 1987, Kufeld et al. 1989). In Montana, emigration rates were high in juvenile

males (16 of 24), low in juvenile females (1 of 29), and occasional in adults. Distances varied

from 11 to 140 km (Wood et al. 1989). Sixty per cent of male and 35% of female yearlings

dispersed in a mark-recovery study in Utah (Robinette 1966). Dispersal distances of black-tailed

deer (Odocoileus hemionus columbianus) in British Columbia and Washington averaged 15.2 km

for males and 12.2 km for females (Bunnell and Harestad 1983).

A study in Pennsylvania documenting dispersal of 308 juvenile male white-tailed found

population densities had no effect on dispersal distance or rate, but that landscape characteristics

affected dispersal distance. Dispersal distance was greatest in areas of least forest cover (Long et

al. 2005). Harvest-induced alterations of the sex ratio within the population seemed to play a role

in the seasonality of dispersal. Although the overall dispersal rate was not affected, autumn

8

dispersal increased when density of adult males increased, and spring dispersal decreased along

with lower density of adult females (Long et al. 2008). The authors also found that spring-time

dispersals were of longer distances than autumn dispersals, and they suggested inbreeding

avoidance behavior required longer distances than did mate competition. Average dispersal

distance was 4.6 km and maximum distance was 8.3 km.

Average dispersal distances in other white-tailed deer studies varied from 3 km in

Virginia (Holzenbein and Marchinton 1992) to 38 km in Illinois (Nixon 1994). The maximum

distance recorded was 212 km in South Dakota (Kernohan et al. 1994). In contrast,

Saskatchewan white-tailed deer emigration distances have been recorded at a mean 215 km and

maximum of 672 km (Stewart and Runge 1985).

1.2.3 Migration

Migration is seasonal movement between non-overlapping ranges. Deer in northern

climates or mountainous areas tend to migrate as an adaptation to cold weather conditions or

changes in seasonal resource availability (Nelson and Mech 1984, Garrott et al. 1987, Sabine et

al. 2002). Migratory deer have potential to spread disease across landscapes. Seasonal home

ranges may be relatively small, but long distances between seasonal ranges may result in

coverage of a large area. Seasonal home ranges of groups of deer may overlap resulting in

increased contact among groups of deer and the potential for long distance disease spread. In

addition, migratory deer may have seasonal ranges in different wildlife or disease management

areas (Brinkman et al. 2005). Migratory behavior of mule deer in Saskatchewan has not been

previously documented and these data provide insight into potential contact routes for disease

transmission.

Migration strategies vary by ecoregion. In mountain foothill regions, mule deer typically

migrate from high summer elevations when forage is abundant, to the protective foothill and

9

basin ranges in winter. At the eastern edge of their range in the prairies, mule deer are usually

non-migratory and inhabit patchy environments along river drainages. The rugged topography is

used as cover to avoid predators and for protection from severe weather (Mackie et al. 2003).

Mixed migration strategies have been observed in local populations and have been attributed to

variable climatic conditions (Nicholson et al. 1997).

Mule deer in a CWD-endemic area of Colorado were much more likely to migrate than

disperse, and the authors suggest that exchange of individuals between population units occurred

most often in the summer months. Only 3 of 151 deer (2%) dispersed, at a distance of 7 to 15

km. The average migratory proportion of the deer studied was 52% but varied between

population units and the average distance was 27.6 km (SE = 1.4) (Conner and Miller 2004). A

previous study in northwest Colorado found 100% of female mule deer migrated an average

distance of 27 km (Garrott et al. 1987). In Wyoming, Sawyer et al. (2005) found a high rate and

distance of migration: 95% of the 166 collared (mostly female) mule deer migrated an average

distance of 84 km (range 20–158 km).

1.2.4 Survival

Survival rates are essential information in population dynamics and management. Causes

of mortality include hunting, predation, disease, malnutrition, winter severity, vehicle collisions,

interspecific competition and habitat loss or change (Wood et al. 1989, White et al. 1987,

Unsworth et al. 1999, DelGiudice et al. 2002). Miller et al. (2008) found that prion-infected deer

(n = 57) were 3.84 times more likely to die (95% CI: 1.64–8.99) than their uninfected

counterparts, and that mortality from mountain lion (Puma concolor) predation was higher than

expected. In order to assess the impact of CWD on Saskatchewan deer populations in the future,

knowledge of present survival trends must be measured for comparison. In addition, seasonal

survival rates provide insight on the effects of the herd reduction program in the study area.

10

This research project was not designed to assess causes of mortality as radio-tracking was

too infrequent. Typically carcasses when found were scavenged and a cause of death could not

be determined. Predators of deer commonly observed in southern Saskatchewan include coyotes

(Canis latrans) and domestic dogs (Canis lupus familiaris). Runge and Wobeser (1975) surveyed

winter mortality of deer in Saskatchewan, and found that when predation was the cause of death,

domestic dogs killed 12 of 26 deer, and coyotes killed 3. Predator species occasionally reported

in the area include mountain lions, bobcats (Lynx rufus), and wolves (Canis lupus), but sightings

are rare.

1.2.4.1 Body condition

Mule deer metabolic rates exhibit an annual cycle corresponding to forage availability.

Summer intake of high quality forage results in mass gains followed by declines during the

winter (Bandy et al. 1970). After 18 months of age, males and females show differences in

seasonal body fat as a result of variation in energetic demands. Loss of body fat is most

pronounced in males, and their reserves are lowest following rut and into the winter and spring;

for females, the low point is following gestation and lactation (Anderson et al. 1972). During the

winter, deer in northern climates reduce their metabolic rate, restrict food intake and activity, and

favor environments suitable to energy conservation. Energetic deficits may result in die-offs

during the winter, but stores of energy depend on nutritional needs being met on the annual range

(Mackie et al. 2003).

In arid environments, precipitation is a key factor in forage quality and droughts can

result in population decline (Lawrence et al. 2004). Bender et al. (2007a) related body condition,

body fat, and precipitation to survival of female mule deer in New Mexico. Body condition in

turn can affect fawn recruitment rates (Wood et al. 1989). Weather patterns, forage quality, and

body condition are crucially interrelated factors affecting population dynamics.

11

1.2.4.2 Home range habitat and survival

Forage availability predicts deer body condition and is related to survival, and recent

studies have assessed home range habitat selection measures in terms of survival. Alaskan black-

tailed deer were found to be at varying risks of mortality depending on the type and stage of

managed forest treatment they used (Farmer et al. 2006). Klaver et al. (2008) found a

relationship between survival and seasonal range characteristics for white-tailed deer in South

Dakota and Wyoming. Survival was highest in ranges with higher proportions of large trees and

with lower proportions of grasses and forbs. Mule deer fawns in Idaho were at greater risk to

predation by coyotes on steep slopes, whereas fawns killed by mountain lions were in areas of

greater cover or structure (Bishop et al. 2005). Caribou mortality through predation was related

to use of forested areas, whereas lowland areas (peatland bogs and fens) provided some

protection from predators (McLoughlin et al. 2005).

Mule deer require rugged terrain and shrubland for cover; grass and shrubland for forage;

and wetlands for forage, cover and water requirements. We tested whether the home range

proportions of these habitats influenced survival.

1.2.4.3 Radio collar effects on survival

Wildlife research often involves animal handling and marking, which may affect the

animal’s survival, behavior, and breeding success. An assumption of survival analyses is that

radio transmitters do not affect survival (Winterstein et al. 2001). Fawns with radio collars were

previously suspected to be at a higher risk of predation, but at least one study found that fawns

with lightweight, inconspicuous ear-tag transmitters were equally at risk (Garrott et al. 1985).

Côté et al. (1998) found that capture negatively affected reproduction and caused kid

abandonment by 3 and 4-year-old mountain goats in Alberta, and found that kids with radio

collars had poorer survival than uncollared kids, although the effect was not statistically

12

significant. Radio collars negatively affected survival of juvenile San Joaquin kit foxes,

particularly when they were no longer under maternal care during dispersal periods (Cypher

1997). The ratio of collar mass to body mass is usually used as a guideline when selecting collars

(White and Garrott 1990) and the Canadian Council on Animal Care (2003) recommends 5% as

a limit. We tested whether exceeding a limit of 2% affected survival.

1.3 Objectives

1. To estimate long distance movements of collared deer in Southern Saskatchewan

including rates and orientation, and build some predictive ability, and to assess the potential of

these movements to increase the geographic range of CWD.

2. To estimate probability of mule deer survival in southern Saskatchewan and estimate the

role of potential factors on survival. Factors were divided into intrinsic and extrinsic measures,

including body measures and condition, and home range features. The impact of radio collar on

survival is also addressed.

This thesis is organized into chapters resembling journal articles. Chapter 2 addresses the

first objective and chapter 3 the second objective. Chapter 4 is a thesis synthesis.

13

CHAPTER 2 DISPERSAL AND MIGRATION OF WILD DEER

IN A CWD-ENDEMIC AREA OF SOUTHERN SASKATCHEWAN

2.1 Abstract

Chronic wasting disease (CWD) is endemic in mule deer (Odocoileus hemionus) of

southern Saskatchewan near the South Saskatchewan (S.S.) River. Exchange of, or contact

between, individuals from endemic areas and naïve subgroups of deer via dispersal or migration

are potential methods of CWD spread. The ability to predict the likelihood and direction of long

distance movements by mule deer from endemic areas will be valuable for disease management.

We radio-collared and tracked 145 mule deer and 19 white-tailed deer from 2006–2009 to

characterize their movement patterns. Results indicated adult migration and juvenile dispersal

distances were similar. Male and female migration distances were also similar. Deer from a study

area of vast grassland were more likely to migrate than deer in study areas fragmented by

agricultural land. Proximity to grassland and high terrain ruggedness values were predictors of

long distance movement paths. Migration and dispersal movements were predicted by terrain and

grassland habitat associated with the S.S. River but mean migration orientation was not predicted

by the S.S. River; rather, it was aligned with the expanse of grassland and hilly terrain of the

Missouri Coteau north of the S.S. River. Observations of white-tailed deer (Odocoileus

virginianus) movements in Saskatchewan and implications for CWD distribution are briefly

discussed. Deer migration and dispersal patterns can be used in conjunction with home range,

contact, and CWD transmission information to predict CWD spread across local landscapes.

14

2.2 Introduction

Chronic wasting disease (CWD), a cervid-specific transmissible spongiform

encephalopathy (TSE), was first detected in a wild mule deer (Odocoileus hemionus) in

Saskatchewan, Canada, in 2000. The provincial Ministry of Environment (MOE) promptly

organized a herd reduction program to reduce deer density and thereby decrease disease

transmission. Increased hunter opportunities were provided in select wildlife management zones

where CWD had been detected. Herd reduction failed to eradicate CWD from the wild, and its

prevalence and geographic extent have steadily increased over the past ten years (CCWHC,

unpublished data).

Chronic wasting disease has multiple routes of transmission and continues to spread in

wild cervid populations, resulting in a challenging management scenario. Transmission can

occur by contact with infected deer or prion-contaminated environmental sources (Mathiason et

al. 2009). Infected deer shed prions in bodily fluids (Mathiason et al. 2006) and carcasses are

also sources of infection (Miller et al. 2004). Prions remain infective in the environment for

periods of years (Johnson et al. 2006, Wiggins 2009), further complicating CWD management.

Host movement patterns likely play a role in wildlife disease spread (Miller et al. 2000).

CWD prevalence is heterogeneous in landscapes where it is found, and is probably influenced by

wild deer distribution and movements (Conner and Miller 2004). Recently, Silbernagel (2010)

identified factors affecting mule deer home range size in this area of Saskatchewan. Male home

ranges were larger than female home ranges, and terrain ruggedness, Shannon’s diversity of

habitat, study site, number of habitat patches, and proportion of cropland all influenced home

range size. Large-scale movements, including dispersal and migration, also affect disease

expansion through time and space in wild cervid populations (Conner et al. 2008). There have

been no previous studies of long distance movements by mule deer in Saskatchewan.

15

Deer movements outside the home range can be long distances during certain life stages.

Dispersal movements are most often by young males and sometimes by young females

(Robinette 1966, Hawkins and Klimstra 1970), whereas dispersal movements by adults are less

common (Wood et al. 1989). Migration is an adaptation to seasonal changes in energetic

demands and resource availability. In mountainous regions, mule deer migrate to take advantage

of quality forage at higher elevations in summer and take cover at lower elevations in winter

(Garrott et al. 1987). Mule deer are generally non-migratory in the prairies (Mackie et al. 2003),

but previous studies have documented mixed strategies (Nicholson et al. 1997). Seasonal

migration results in varied deer density and association rates—deer typically congregate in

winter yards where they contact each other or share habitat more frequently (Conner et al. 2008,

Silbernagel 2010). Thus, migration may influence disease transmission rates whereas dispersal is

thought to affect disease spread to new areas (Conner et al. 2008). Even short-term movements

outside the home range have potential to transmit disease (Skuldt et al. 2008).

Mule deer studies in the Canadian prairies are scarce and disease management decisions

have been made without scientific knowledge of local deer movement behavior. We attempted to

fill this knowledge gap by radio-collaring and tracking wild deer in a CWD-endemic area. Our

objective was to estimate distance of movements by radio-collared mule deer in Southern

Saskatchewan by determining rates and orientations of dispersal, migratory, and excursion

movements. We examined landscape factors associated with long distance movements, and

assessed the potential of these movements to further increase the geographic range of CWD in

Saskatchewan.

16

2.3 Methods

2.3.1 Study Area

The study was conducted along the South Saskatchewan River basin in the Prairie

Ecozone of southern Saskatchewan, Canada. The landscape was dominated by relatively flat

low-lying terrain interrupted by areas of kettle topography including river valleys and coulees,

sand hills, and rolling terrain.

The total study area size was 2740 km2 and included 5 study sites: Antelope Creek (ANT;

248 km2; 50.66°N, 108.27°W at center), Beechy pasture (BEE; 613 km2; 50.98°N, 107.70°W),

Douglas park (DOU; 605 km2; 51.02°N, 106.44°W), Matador pasture (MAT; 810 km2; 50.77°N,

107.73°W), and Swift Current Creek (SWI; 464 km2; 50.58°N, 107.73°W). See Figure 2.1.

The study areas were within two prairie ecoregions: Mixed Grassland and Moist Mixed

Grassland. Short grasses in the area were blue grama (Bouteloua gracilis) and sedge (Carex

spp.) and mid-to-tall grasses included wheatgrasses (Agropyron spp.), Junegrass (Koelaria

macrantha), needle-and-thread (Stipa comata), and western porcupine (Stipa spartea v.

curtiseta). Pasture sage (Artemisia frigida) and moss phlox (Phlox hoodii) were common forbs;

common shrubs included Western snowberry (Symphoricarpos occidentalis), wolf willow

(Eleagnus commutata), creeping juniper (Juniperus horizontalis), chokecherry (Prunus

virginiana) and Saskatoon berry (Amelanchier alnifolia) (Acton et al. 1998). Trees were

uncommon except in the DOU site, where the dominant species were trembling aspen (Populus

tremuloides) and balsam poplar (P. balsamifera). Common crop types in the grassland region of

Saskatchewan included wheat, barley, oats, canola, mustard, peas, lentils, and flax (Government

of Saskatchewan 2009).

Saskatchewan has long, cold winters and short, warm summers. Average temperature in

the south during the coldest month of winter (January) was -13°C and in the warmest month of

17

summer (July) was 19°C. Mean annual precipitation was 414 mm (Canadian Plains Research

Center 2006).

The total area and proportions of grassland and cropland varied between sites (Figure

2.2). ANT and SWI sites were dominated by agricultural land (Figure 2.3) parceled by the public

land survey system with regular grid roads every 2 by 1 mile. Core areas of deer use existed in

the rugged grassland terrain along the S.S. River or its drainage streams. DOU was named after

Douglas Provincial Park in the core of its study area; deer within park boundaries were afforded

protection from hunting pressure which helped maintain high deer densities. DOU, BEE, and

MAT sites were characterized by large expanses of hilly native pasture with agricultural areas at

the periphery. These areas had sparse human habitation and few roads. All areas except BEE

were adjacent to the S. S. River which was dammed near the DOU site forming Lake

Diefenbaker. Although BEE and MAT sites were considered independent at the beginning of the

study, early observations indicated many deer moved seasonally between the two sites and for

analyses we considered them as one study site (hereafter referred to as BEEMAT).

Figure 2.1: Map of study sites in southern Saskatchewan along the South Saskatchewan River. WMZ are wildlife management zones.

18

19

Figure 2.2: Total landcover area for each study site.

Figure 2.3: Proportional landcover for each study site.

20

CWD has been found in wild deer in all study areas except DOU. Population reduction

programs to manage CWD began in SWI in 2002, expanded to include ANT, and MAT the

following year, and in 2007 expanded to include BEE. White-tailed and mule deer ranges

overlap in Saskatchewan and both species were found in the study areas. Mule deer were more

common and had a higher prevalence of CWD. Figure 2.4 shows the distribution of the total

number of CWD-positive wild deer detected in the province (Saskatchewan Ministry of

Environment 2010). The study site was centered on the area north of Swift Current (Figure 2.5).

Figure 2.4: Locations of CWD positive wild deer in Saskatchewan, updated January 14, 2010.

Numbers indicate wildlife management zones.

21

Figure 2.5: Inset of previous map showing study area (dashed lines). Numbers indicate wildlife management zones.

2.3.2 Capture

Between April 2006 and April 2008, mule and white-tailed deer were captured by

helicopter net-gunning (Bighorn Helicopters Inc., Cranbrook B.C.) (Krausman et al. 1985) and

Clover traps (Clover 1956). Deer caught in traps were anaesthetized immediately after

technicians approached and collapsed the trap, and once anaesthetized, blindfolds and hobbles

were applied. Deer transported by helicopter were blindfolded and hobbled prior to

anaesthetization. Each deer was anaesthetized using Xylazine-Telazol (Rompun®; Telazol®) in

order to test for CWD on tonsil biopsy (Schuler et al. 2005), and later reversed with Atipamezole

(Antisedan®). Age was assessed by body size and tooth wear (Severinghaus 1949, Robinette et

al. 1957). Deer aged 6 to 10 months old at capture were classed as juveniles; between 1−2 years

as yearlings; and over 2 years as adults. Individuals were ear-tagged with a numbered plastic

cattle tag color-coded to study site, and a small metal numbered ear-tag (Ketchum Manufacturing

22

Inc., Brockville ON K6V 7N5, Canada). Radio collars were Lotek GPS 3300 or 4400 or VHF

(Lotek Wireless Inc., Newmarket ON L3Y 7B5, Canada) or expandable VHF (Advanced

Telemetry Systems, Isanti, MN 55040, USA). Capture protocol was approved by the University

of Saskatchewan Animal Care Committee (20050135).

In 2008 we attempted to reduce collar loss from slippage on juveniles. Males were

collared with expandable VHF and females with fixed-circumference VHF and a sample (n = 4;

1F, 3M) were collared with a lightweight (300 g) GPS transmitter (Televilt by Followit Holdings

AB, Lindesberg, Sweden). The latter group’s collars failed prematurely. In all capture years, all

non-expandable male collars were fitted with a nylon-enveloped foam insert to allow for neck

expansion during the rut. Collars had mortality sensors activated by a period of immobility (6 to

12 hours, depending on the collar type).

GPS collars were set to collect locations every 4 hours, or every hour during breeding and

fawning seasons, and also recorded altitude and temperature. Minimum monthly locations (VHF)

or signal checks (GPS) were acquired via fixed-wing aircraft telemetry (Mech 1983) or on the

ground by hand-held antenna telemetry. Location accuracy for VHF collars, estimated from

known collar locations, was 678 m (SE = 48, n = 82). GPS collar accuracy was reported by

LOTEK at 5 m. We evaluated their accuracy at 11.3 m (SE = 2.6, n = 4). To assess accuracy, we

calculated the mean x and y positions of stationary collar data and then calculated the average

distance of all fix locations from the mean. Data were plotted in UTM coordinates for all

analyses.

2.3.3 Definition and Measurement of Long Distance Movements

Dispersal was defined as a permanent movement from an animal’s natal range to a new,

non-overlapping range. Dispersal is usually undertaken by juvenile males, between 6 and 24

months of age, but has been reported in all age-sex classes (Robinette 1966, Holzenbein and

23

Marchinton 1992, Nelson and Mech 1992, Kenward et al. 2001, Nixon et al. 2007). Since the

natal range of only the former group is known, only they were included in dispersal observations.

Migration was defined as movement between non-overlapping seasonal ranges (Brown 1992,

Brinkman et al. 2005, Sawyer et al. 2005). Obligatory migrants moved to winter range in early

winter where they resided until spring (Sabine et al. 2002), whereas conditional migrants failed

to migrate during one season or migrated unpredictably (Nelson 1995, Brinkman et al. 2005).

Resident deer had one home range area year-round and never migrated (Vercauteren and

Hygnstrom 1998). Short-term movements outside the normal home range area (i.e., <1 month)

were termed “excursions,” as long as the deer returned to its normal range. These were observed

for all categories of deer.

Home range polygons were calculated as 95% kernel density estimates (KDE) (Rodgers

et al. 2007) in ArcGIS 9.2 (Environmental Systems Research Institute, Redlands, Calif.). Least

squares cross validation (LSCV) is recommended for smoothing factor (h) selection in KDE;

however, LSCV methods will generally fail at sample sizes > 100 or < 10 (Hemson et al. 2005).

We used LSCV methods for VHF-collared deer with suitable sample sizes. GPS-collared

individuals had sample sizes > 100 and for these 275 m was selected as a smoothing factor

because href (software-generated reference bandwidth) tends to over-smooth and inflate home

range size (Seaman et al. 1999). The goal was to delineate separate ranges objectively. Home

range size was not evaluated in this study.

The centroid of each range polygon was calculated (ET SpatialTechniques, Pretoria,

South Africa) and used to measure movement orientation and distances between seasonal or

natal and adult ranges (Kernohan et al. 1994, Zar 1999). For seasonal ranges, we measured the

travel orientation from winter range to summer range. For excursions and atypical migrations,

24

we measured the travel orientation from the dominant range (where it ranged the majority of its

time) to the short-term range. For dispersing deer, we measured the orientation from natal range

to adult range. Distances between centroids <5 km were not considered long distance

movements. Date of movement was recorded as the first location outside of the previous range,

or the first date the individual was not located in its usual range if it was subsequently located in

a new range. GPS data were accurate to the day and VHF data were accurate to the month.

We removed deer from analyses if their research lifespan was shorter than 6 months due

to collar loss, collar failure, or death. Exceptions were made for 2 juvenile deer that died within

6 months of capture but had clearly dispersed. We excluded resident deer studied less than a year

from analyses because they had potential to migrate but died. We classified migratory deer

studied for 18 months or longer as conditional or obligatory.

2.3.4 Dispersal Rate

Since dispersal rate calculated as the number of dispersals per juvenile captured would be

inaccurate due to death loss and new captures varying the number of juveniles available at each

dispersal period, we estimated annual dispersal rate for each cohort using an adaptation of

Kaplan-Meier survival model (Holzenbein and Marchinton 1992, Kaplan and Meier 1958,

Pollock et al. 1989, Long 2006). In a survival model, deaths reduce the population survival rate,

whereas in this model, dispersals reduce the philopatry rate. Dispersal rate is simply one minus

the philopatry rate. Rates were calculated over monthly intervals and mortalities were censored

from the number of individuals available to disperse.

2.3.5 Factors Associated with Long Distance Movements

The movement paths of a subset of GPS-collared mule deer (n = 33) that migrated or

dispersed were analyzed to determine features associated with locations along the chosen travel

25

path between home ranges. These features may describe habitat preferences of travelling mule

deer and can be used to predict deer movement and future disease spread.

We used the last location in the home range, first location in the new home range, and all

locations between to represent the chosen travel path. A minimum of 3 GPS positions (collected

in 1, 4, or 6-hour intervals) was used to create a digitized line representing the travel path. Using

alternate animal movement routes tool (Jenness 2005) in ArcView® GIS 3.2 (Environmental

Systems Research Institute, Redlands, Calif.), ten lines of equal length and shape were randomly

generated from the point of origin for each deer’s path (Long 2006) to represent alternatives to

the selected path. Vertices of all lines were used as sample points (n = 4796) to measure

landscape attributes (Bruggeman et al. 2007) including habitat (land cover type, patch size and

patch area–perimeter ratio), topography (TRI and elevation), and proximity to landscape features

(river, wetland or open water, paved or grid roads, grassland and cropland). Because the point of

origin was equivalent for chosen and random lines, it was removed from analysis.

A 56m-resolution 2006 land cover map was acquired from the Agricultural Financial

Services Corporation (AFSC; Lacombe, AB; T4L 1B1). We simplified the 9-class cover map

into 7 classes: annual cropland, grassland (pasture), forage (hay), forest, wetland, water, and

other (built-up, barren, or unclassified). Shrubland was not a class in the cover map but most

grassland in the study area is native and partly shrub-covered. Digital elevation model data were

transformed into a terrain ruggedness index (TRI) (Russell and Levitin 1995). Proximity to

landscape features was calculated with the near tool in ArcInfo 9.3 (Environmental Systems

Research Institute, Redlands, Calif.).

2.3.6 Statistical Analyses

Statistics on migration were calculated using only adult deer and dispersal using only

deer captured as juveniles. Data distributions for each category of movement were tested for

26

normality. Data distributions tested normal (1-sample KS test: excursions, P = 0.32; migrations,

P = 0.077; dispersals, P = 0.087), but migration and dispersal distances were tested with non-

parametric methods because they were approximately non-parametric (Figure 2.6). For

migrations and dispersals, differences in movement distances between sexes were evaluated with

a Mann-Whitney U test, and for study sites a chi-square test. For excursions, we used t-tests for

differences between sexes and one-way ANOVA for study site differences. The low sample size

of dispersals prevented statistical comparisons between sexes. All statistical analyses were

performed with SPSS software (version 17.0; SPSS Inc., Chicago, Ill.).

Figure 2.6: Frequency histograms of dispersal distance (left) and migration distance (right).

Directional analyses of migration and excursion datasets were completed using Oriana

software (Kovach Computing Services©). For data independence, we selected one migration per

deer (winter to summer) and randomly selected one excursion per deer. Two circular

distribution tests assessed directionality: Rayleigh’s uniformity test and Rao’s spacing test.

27

Rayleigh’s uniformity test detected normal distribution of directions around a circle, and if the

null hypothesis was rejected, then it was most likely that a direction was preferred. Rao’s

spacing test also assessed normality of directions, where the null hypothesis was that deer

movement directions were random and distributed uniformly about a circular compass (0 to

360°). Rao’s test was considered a stronger test because it assessed whether the distribution was

evenly spaced, i.e., spacing between points should be approximately 360°/n. The latter test

detected clusters of directionality whereas Rayleigh’s might not (Kovach 2009).

The landscape features associated with chosen paths for long distance movements were

assessed using generalized estimating equations (GEE) with an exchangeable correlation matrix,

binomial distribution, and logit link function. Data were clustered by individual deer as subjects,

and further clustered within each subject by unique path and sequential location along path.

Variables that scored P < 0.2 in simple regression analyses were included in multiple regression

models. Backward removal of variables was then performed with significance for model

inclusion set at α ≤ 0.05. Because GEE are non-likelihood based, QIC (quasi-likelihood under

independence model criterion) was recommended over AIC for GEE model selection (Pan

2001). QIC values were assessed to select the correlation structure and QICC (corrected quasi-

likelihood under independence criterion) values were assessed to select the best set of parameters

(Garson 2009). Final model variables were tested for correlation and interaction with one

another. If Spearman’s r was > 0.7, one of the correlated variables was removed based on

biological reasoning and/or based on significance values.

28

2.4 Results

2.4.1 Long Distance Movements

One hundred and sixty-four deer (145 mule, 19 white-tailed) were used to categorize

movement. Of these, 4 lived less than a year but 141 were studied for a period of 1 to 3 years,

and the average time in the study was 21 months. Fourteen juveniles were classified as

dispersers, 55 adults as migrants, and 76 adults as residents.

2.4.1.1 Dispersal

Between spring 2007 and autumn 2008, 14 juvenile mule deer dispersed a median

distance of 22.8 km (SE = 13.6; x̄ = 39.9 km; range 6.5−195.5 km). More males (n = 10)

dispersed than females (n = 4) and 11 (4 F, 7 M; 5 in 2007, 6 in 2008) dispersals were initiated

in the spring around fawning time (June). Three dispersals occurred during autumn rut; all were

male and all occurred in 2007. Ten deer were approximately 12 months of age at time of

dispersal (7 M, 3 F), 3 were 18 months (M), and 1 was 24 months (F). Most dispersals were from

ANT (n = 8), and 2 each were from the other sites (Table 2.1). Pearson’s χ2 results (1.145, df =

3, P = 0.766) showed no difference between study site in the proportion of dispersers and

residents.

29

Table 2.1: Mule deer dispersal details

deer_id site

dispersal

distance

(km)

bearing

(°N) season age

(months) sex Dispersal

initiation date capture date

O016 ANT 54.7 101 spring 12 F 18-May-07 13-Mar-07

Y024 BEEMAT 17.5 19 spring 12 M 29-May-07 12-Mar-07

B028 DOU 98.9 27 spring 13 F 05-Jul-07* 16-Mar-07

O022 ANT 33.6 99 spring 13 M 05-Jul-07* 13-Mar-07

O050 ANT 25.0 320 spring 13 F 05-Jul-07* 09-May-07

O010 ANT 6.5 105 autumn 17 M 01-Nov-07 13-Mar-07

O011 ANT 20.6 284 autumn 17 M 06-Nov-07* 13-Mar-07

P049 SWI 8.4 105 autumn 18 M 06-Dec-07* 14-Mar-07

Y069 BEEMAT 15.3 332 spring 11 M 06-May-08* 09-Mar-08

O057 ANT 33.2 93 spring 13 M 14-Jun-08 09-Apr-08

O053 ANT 25.0 107 spring 13 M 21-Jun-08 22-Mar-08

B027 DOU 11.3 95 spring 24 F 25-Jun-08* 16-Mar-07

P053 SWI 195.5 283 spring 13 M 25-Jun-08* 08-Mar-08

O059 ANT 13.2 335 spring 14 M 05-Aug-08* 12-Apr-08

*indicates flight date--dispersal date is approximate

We intended to compare study site dispersals by collaring an equal number and similar

gender ratio of juveniles per site, but were unsuccessful due to mortality, collar losses, and

difficulty finding juveniles in some areas. As a result, the number of juveniles studied at each site

was unequal (Table 2.2). Nevertheless, we have reported the results from a study site context.

Juvenile mule deer included in analyses for each site were 15 (ANT), 9 (BEEMAT), 12 (DOU),

and 8 (SWI). ANT had a higher number of juvenile males collared than any other site. Numbers

of juveniles available (alive and had not dispersed previously) per dispersal period averaged over

the 4 seasonal dispersal opportunities were 5.7 (ANT), 3.6 (BEEMAT), 4.2 (DOU), 4.5 (SWI).

In June 2007 there were 10 juveniles available at ANT, compared to 4, 5, and 6 respectively at

BEEMAT, DOU, and SWI. This was the first potential dispersal period of the study and also had

the highest contrast in number available per site.

30

Table 2.2: Summary of juveniles and dispersal movements by study site.

ANT BEEMAT DOU SWI ALL

Captured juveniles (JUV) 15 9 12 8 44

male 9 5 4 4 22

female 6 4 8 4 22

Average # JUV available 5.7 3.6 4.2 4.5 4.5

male 2.3 0.5 2.0 2.0 1.7

female 3.5 3.2 2.2 2.5 2.8

Dispersal count 8 2 2 2 14

male 6 2 - 2 10

female 2 - 2 - 4

Avg. dispersal distance (km) 26.5 16.4 55.1 102.0 39.9

male 22.0 16.4 - 102.0 36.9

female 39.8 - 55.1 - 47.5

Using the dispersal rate adapted from Kaplan-Meier, the dispersal rate from capture in

late winter until year-end for the 2007 cohort was 0.36 (±0.21), and for the 2008 cohort was 0.65

(±0.32) (Figure 2.7). The overall dispersal rate from spring 2007 through autumn 2008 with both

cohorts was 0.55 (±0.17) (Figure 2.8). There were no dispersal events in autumn 2008, the last

dispersal period studied.

The farthest dispersal distance of 195 km was between capture at Swift Current Creek

and mortality location in southeastern Alberta along the South Saskatchewan River. Dispersing

deer from ANT and SWI tended to travel with the orientation of the river, as seen in Figure 2.9.

Two dispersers from ANT settled in SWI (O016 and O057) and another (O016) dispersed to

SWI but died during settlement 2 weeks after leaving its natal range. One juvenile white-tailed

dispersed—a male from SWI that moved 36 km to ANT at 12 months of age and was dead where

it was found.

31

Figure 2.7: Annual juvenile dispersal rate for 2007 and 2008 cohorts.

Figure 2.8 Overall juvenile dispersal rate 2007–2008. Confidence intervals vary through time depending on the number of juveniles alive in the study, narrowing with additional captures and widening with losses.

32

Figure 2.9: Juvenile dispersals from the ANT site and 1 from SWI to ANT. All are mule deer except P035, a white-tailed deer from SWI.

2.4.1.2 Migration

Median migration distance of adult mule deer was 16.0 km (SE = 2.6, x̄ = 22.8 km) and

ranged from 5.0 to 112.6 km. Forty-two per cent of adults were migratory (n = 49) and 58%

were resident (n = 67). Most of the migratory deer were from the BEEMAT site, where 68% of

adults migrated compared to 24 to 30% at other sites (Table 2.3). The proportion of migratory

deer per study site differed significantly (χ32 = 19.533, P ≤ 0.001). However, relative proportions

could have been inflated because deer that behaved as residents but lived less than 1 year (n = 6)

were removed from analyses, whereas deer that migrated and lived between 6–11 months were

33

included (n = 13). Migration distances were not equivalent between study sites (χ32 = 9.392, P =

0.025). Tested pair-wise, we found significant differences between BEEMAT and DOU (Z1 =

-2.105, P = 0.039) and BEEMAT and SWI (Z1 = -3.040, P = 0.002).

Thirty-nine per cent of males and 45% of females were migratory (Table 2.3). Median

migration distance was similar for both sexes (Table 2.4) (Z1 = -0.357, P = 0.721).

Table 2.3: Number and proportion of migratory and resident adult mule deer, by study site and by sex.

migratory resident Total ADULT MULE n % n % n

SITE ANT 6 24% 19 76% 25 BEEMAT 30 68% 14 32% 44 DOU 5 25% 15 75% 20 SWI 8 30% 19 70% 27 SEX F 29 45% 36 55% 65 M 20 39% 31 61% 51 Total 49 42% 67 58% 116

Table 2.4: Adult mule deer migration distance by study site and by sex.

distance (km) ADULT MULE

median SE x̄ Min max n

SITE ANT 16.0 4.9 18.3 5.6 37.6 6 BEEMAT 25.8 1.9 24.2 8.1 48.4 30 DOU 12.5 1.7 11.2 6.0 15.6 5 SWI 8.9 14.0 28.2 5.0 112.6 8 SEX F 19.7 3.7 23.5 5.6 112.6 29 M 15.7 3.5 21.7 5.0 65.4 20

Of 29 migratory deer studied a minimum of 18 months, 16 (55%) were conditional and

13 (45%) were obligate migrants. The latter category included 8 from BEEMAT, 2 from DOU,

and 3 from SWI. The gender ratio of conditional migrants was similar (7 F, 9 M), whereas

obligatory migrants were most often female (10 F, 3 M). Median migration distance was similar

between obligatory (16.0 km) and conditional (14.7 km) migrants (Z1 = 0.219, P = 0.846).

34

At least one repeat movement did not fit into our description of obligate migration. A

male from DOU made a 25 km migration for approximately 1 month (August) in 2 consecutive

years: this short-term movement did not fit into the regular seasonal pattern so was not

considered obligate.

2.4.1.3 Dispersal vs. migration

Mule deer migration distances were similar to dispersal distances (Z1 = -1.134, P = 0.26).

Five movements by adult deer fit the description of dispersal but were more appropriately

classed as conditional migrations because their natal ranges were unknown. They were marked

by a long distance movement preceded by annual residence in an established range or followed

by residence in the new range for at least 2 seasons. The longest distance of an adult conditional

migrant was that of a 3-year-old GPS-collared mule doe (P012) that was captured February 2007

in the SWI study site. She held a home range in the area for the next 16 months until she

departed June 8, 2008 to a destination 113 km away in a south-western direction (253°). She

stayed in a small home range area for approximately one month, presumably fawning, and then

relocated to another range 10 km to the east until September 6, 2008. For the next month, GPS

locations were sporadic but she was headed eastward and died of unknown cause October 8, 53

km east of the fawning range.

Average migration distance by white-tailed deer was 37.6 km (n = 6; SE = 8.4 km; range

9.2–79.8 km; 4 F, 2M). June 6–8, 2007 an adult white-tailed doe (B063) moved 80 km from her

capture location in DOU at 3 years of age. Her destination to the northeast was similar to that of

a female mule deer that dispersed in 2007 (B028). A year later B063 made a brief visit to her

original home range from June 13 to July 1, 2009 and then returned to her new range.

35

2.4.1.4 Excursions

Mean excursion distance was 9.0 km (SE = 0.8, n = 56 deer). Average distance for

females was 8.6 km (SE = 1.0, n = 38) and for males was 9.8 km (SE = 1.1, n = 18). Excursion

distances did not differ between sexes (t55 = -0.745, P = 0.459) or study sites (F3, 55 = 1.214, P >

0.3) (Table 2.5). Mean excursion distance was similar for all months (F 11, 55 = 1.059, P > 0.4)

(Table 2.6 and Figure 2.10).

Table 2.5: Excursion distance (km) by study site

site x̄ SE Median Range n

ANT 9.0 1.8 6.4 18.9 13

BEEMAT 10.4 1.2 8.9 30.1 25

DOU 7.2 1.5 6.5 12.9 10

SWI 6.8 1.7 4.2 12.5 8

Total 9.0 0.8 7.8 32.0 56

Table 2.6: Mean excursion distance (km) per month

Month n of deer

Mean distance

(km) SE

Jan 8 10.5 3.7 Feb 3 8.5 3.4 Mar 14 9.1 1.1 Apr 10 7.1 1.6 May 6 11.5 2.0 Jun 6 11.7 2.6 Jul 3 4.4 0.2

Aug 9 6.5 0.8 Sep 3 7.0 1.5 Oct 7 6.1 0.9 Nov 16 8.0 1.3 Dec 11 9.3 1.3

Annual 96 8.5 0.5

36

Figure 2.10: Mean excursion distance (km) per month

Excursions by males were most common during peak breeding season in November

(Figure 2.11). Female excursions were least common in the summer months when fawns were

young, but were a regular occurrence the rest of the year.

37

Figure 2.11: Count of excursions made by mule deer for each month. All years of data were pooled.

Most excursions were of short duration (Table 2.7). Female movements outside the range

were most often less than a day and male were 1 day (Figure 2.12). Duration was significantly

different between sexes (Z1 = -4.368, P ≤ 0.001).

Table 2.7: Period (days) of excursion by sex.

sex Mean n Median SE Range

F 2.62 133 .00 .569 31

M 4.41 64 1.00 .983 27

Total 3.20 197 .00 .502 31

38

Figure 2.12: Period (days) of mule deer excursions by sex, showing non-normal distribution.

The South Saskatchewan River appeared to be somewhat of a barrier to dispersal and

migration movements, but not to all deer. Six individuals crossed the S.S. River one time, an

additional 4 crossed twice, and one individual crossed 4 times. Five were from ANT, 4 from

SWI, and 1 each from MAT and DOU. River crossings occurred at various times of year. Within

the study site, the S.S. River is approximately 1 km wide.

39

2.4.2 Attributes of Long Distance Movements

2.4.2.1 Travel speed

The average travel speed of dispersal was calculated from 5 GPS-collared dispersing

juvenile deer at 769 m/hr (SD= 506; range: 1–2819 m/hr). Three of these juveniles dispersed

over a short period (3–48 hours) and 2 juveniles dispersed over a longer period (9–15 days) that

involved some days of rest (Table 2.8).

Table 2.8: Travel speed and duration of dispersal for GPS-collared mule deer.

deer_ID Average

m/hr

Max

m/hr

Min

m/hr

Elapsed

hr

Fix

interval

(hr)

Days

O010 1420 1697 1250 3 1 0.1

O016 187 1238 1 356 4 14.8

O053 775 1595 151 48 6 2

O057 373 2819 24 224 6 9.3

Y024 1092 2706 77 28 1 1.2

overall 769 2011 301 132 4 5.5

SE 226 317 239 68 1 2.8

SD 506 708 534 153 3 6

The average travel speed of migrating mule deer calculated from 28 GPS-collared adults

was 1039 m/hr (SD=366; range 5–4162 m/hr). Most migrations took place in less than a day

(Table 2.9).

40

Table 2.9: Travel speed and duration of migration for GPS-collared mule deer.

deer_ID Average

m/hr

Max

m/hr

Min

m/hr

Elapsed

hr

Fix

interval

(hr)

Days

B022 807 1539 76 4 4 0.2

B029 710 2101 77 52 4 2.2

B049 727 1492 63 20 4 0.8

B050 1260 1993 527 8 4 0.3

B064 623 1267 285 28 4 1.2

G030 1052 1238 865 8 4 0.3

G032 721 1230 178 12 4 0.5

G034 779 2004 142 104 4 4.3

O003 366 436 291 12 4 0.5

O045 1490 3505 130 30 1 1.3

O055 637 932 342 16 6 0.7

P042 1170 1789 300 5 1 0.2

P045 1506 1992 1137 4 1 0.2

P051 1258 1666 851 2 1 0.1

Y022 1088 1921 609 12 4 0.5

Y026 1436 2391 480 8 4 0.3

Y057 1118 1254 983 8 4 0.3

Y060 1299 2786 433 12 4 0.5

Y063 1200 3117 233 16 4 0.7

Y064 734 2371 71 52 4 2.2

Y066 598 1718 15 56 4 2.3

Y067 760 3226 26 52 4 2.2

Y076 1081 2148 15 4 4 0.2

Y078 809 1881 5 40 4 1.7

Y058 1434 2417 133 8 1 0.3

P012 1932 4162 137 108 1 4.5

P041 991 2528 24 81 1 3.4

P052 1517 2674 791 7 1 0.3

overall 1039 2064 329 27 3 1.1

SE 69 154 62 6 0 0.2

SD 366 817 330 30 1 1

41

2.4.2.2 Travel orientation

Mean dispersal direction was 49° (SD= 79°, median = 60°, 95% CI: 351°–108°).

Rayleigh’s uniformity test was not significant (Z = 2.138, P = 0.117, n = 14) but Rao’s test

results indicated spacing was not uniform (U = 219.14, P < 0.01, n = 14). Figure 2.13 shows

directional tendencies around the compass. The southwest quadrant appears to have been

avoided. From DOU or MAT, southwest travel would require deer to cross the S.S. River; from

ANT, BEE, or SWI, southwest travel would lead into land dominated by agricultural activity

with little suitable habitat.

Mean vector of migrations was 41° (SD = 92°; median = 47°, 95% CI: 2°–79°).

Migration direction distribution was not uniform (Rayleigh’s Z = 4.137, P = 0.016, n = 55), nor

was spacing uniform (Rao’s U = 177.7, P < 0.001, n = 55). Study site migration bearings were

not significantly directional except for BEEMAT (Rayleigh’s Z = 6.9; P < 0.001; Rao’s U =

194.6, P < 0.01; n = 33). Mean migration vector at this site was 35° (SD = 72°; median = 27°;

95% CI: 6°–63°) (Table 2.10). Sample sizes for the other sites were low, meaning these

directionality tests have limited use, and the overall mean direction was strongly influenced by

migratory deer in BEEMAT. Direction of travel was randomly and evenly distributed for

excursions (Rayleigh’s Z = 0.26, P > 0.7; Rao’s U = 132.87, 0.50 > P > 0.10; n = 56).

Table 2.10: Mean migration vector by study site. Only BEEMAT had significant directionality.

site DOU BEEMAT ANT SWI

n 6 33 6 10

Mean Vector (µ) 231° 35° 103° 235°

Median 180° 27° 86° 291°

SD of Mean 128° 72° 86° 148°

***** 6° ***** 95% CI (-/+) for µ

***** 63° ***** ***** indicates that a result could not be calculated.

42

site

DOU

BEEMAT

ANT

SWI

N

E

S

W 1000 1000

1000

1000

100 100

100

100

10 10

10

10

Figure 2.13: Distance and direction (km) of dispersal events. Distance scale is logarithmic. The black line on the compass represents overall mean direction (49°) with 95% confidence interval shown in red. Directionality was not significant but directions were significantly clustered about the compass.

43

site

DOU

BEEMAT

ANT

SWI

N

E

S

W 1000 1000

1000

1000

100 100

100

100

10 10

10

10

Figure 2.14: Distance and direction (km) of migration events (1 per deer). Distance scale is logarithmic. The black line on the compass represents overall mean direction (41°) with 95% confidence interval.

44

site

DOU

BEEMAT

ANT

SWI

N

E

S

W 40 40

40

40

30 30

30

30

20 20

20

20

10 10

10

10

Figure 2.15: Distance and direction (km) of excursion events (1 randomly selected per deer). The black line on the compass represents overall mean direction with 95% confidence interval shown in red. Direction of movement was not significant for excursions.

45

2.4.2.3 Habitat selection

The most parsimonious model for chosen travel path on long distance movements

included selection of habitats with rugged terrain and in close proximity to grassland. Terrain

ruggedness held the strongest influence. We expected an autoregressive structure would best fit

the data, but QIC values indicated an exchangeable structure was most appropriate. Competing

models included proximity to water and grid road, but their regression coefficients approximated

zero and QICC values on these models suggest the main model is the only appropriate fit to the

data (Table 2.11).

Table 2.11: GEE model results

Model

1 2

Interval Interval

Parameter P β lower upper P β lower upper

TRI1 0.001 0.190 0.077 0.302 0.000 0.229 0.110 0.347

Grass dist2 0.000 -0.002 -0.003 -0.001 - - - -

correlation structure

EXCH EXCH

QIC 2558.42 2634.55

QICC 2509.20 2590.30

∆QICC 0.00 81.11

Note: parameter meanings: 1. TRI = Terrain ruggedness index, 2. Grass dist = distance to nearest grassland

2.5 Discussion

2.5.1 Long Distance Movement Demographics

Dispersal and migration occurred frequently in our study population, and there was no

significant difference in mean distance moved for the two movement types. The longest

dispersal distance was 195 km by a juvenile male and the longest migration distance was 113 km

by an adult female. For comparison, the longest mule deer dispersal reported in south-central

46

Washington was 113 km (Hedlund 1975) and longest mule deer migration in a mountainous

region of western Wyoming was 158 km (Sawyer et al. 2005). A number of adult deer made

unidirectional long-distance movements, suggesting adult dispersal may be more common than

previously thought. Most migratory deer were from the BEEMAT site where there were large

uninterrupted areas of native grassland.

Due to concerns that herd reduction programs were resulting in a shift in population

structure toward yearling males and a subsequent increase in dispersal rates, we originally

intended to contrast dispersal distances and rates within and outside CWD management areas.

Initially the study design included 2 study sites in wildlife management zones (WMZ) outside of

CWD herd-reduction zones (BEE and DOU) and 2 sites within (MAT and SWI). In the study’s

pilot year, we determined mule deer commonly migrated between BEE and MAT and the BEE

area was subsequently included in the herd reduction zone. ANT was added as a study site in

2007; it too was within the herd reduction zone. Due to changing boundaries of herd reduction

zones and low sample sizes of juvenile deer the goal of contrasting dispersal rates between herd

reduction zones and non-herd reduction zones was not met.

Long distance movements were made by adults, as well as juveniles, and there was no

difference between migration and dispersal rates or distances. Consequently dispersal and

migration movements by mule deer have similar potential for spreading disease across

landscapes. However, since adult male mule deer have higher CWD prevalence (Grear et al.

2006), their movement patterns may pose a greater risk for CWD spread.

2.5.2 Habitat and Long Distance Movements

Habitat composition on the landscape seemed to predict movement frequency and

direction. Migrations occurred less frequently and were of shorter distances in fragmented,

agricultural landscapes. Orientation of migration was predictable in BEEMAT only; otherwise,

47

deer tended to follow the orientation of the S.S. River. Migration strategies vary among mule

deer within their range in North America; in mountainous areas most are migratory, in the

prairies, some are migratory. Although theories on migration strategies differ, Mackie et al.

(2003) succinctly stated “movement patterns employed by individual deer depended on the

spatial arrangement of habitat components they attempted to exploit.”

Juvenile dispersals were frequent on the south side of the S.S. River but the low sample

size of juveniles from BEEMAT limited the confidence in prediction of dispersal rates from a

less fragmented grassland habitat. A study of white-tailed deer found that dispersal distance

increased in habitats with less forest cover, but that dispersal rates were independent of forest

cover and population density (Long et al. 2005). Source-sink models and island biogeography

theory predict that patches of ideal habitat separated by large distances are less likely to be

colonized (Noss and Csuti 1997). Landscape characteristics between fragments of suitable

habitat are important in the dispersal process; however, their effects are poorly understood and

are likely species-specific (Wiens 2001). The amount and arrangement of ideal rugged grassland

habitat within a landscape may predict mule deer dispersal distance, but a larger sample size is

required in order to make this conclusion. Habitat preference and habitat availability in a

landscape can easily be measured and there is increasing evidence landscape can be used to

predict dispersal orientation.

In this study, like many others, observations on dispersal consisted of only x, y locations

and lacked measurement of behavioral cues or other factors that may contribute to a dispersal

event. Research on factors influencing dispersal rate is a formidable task given the cost of

acquiring an appropriate sample size and measuring multiple demographic, landscape, and social

48

parameters. In this study, similar to others, males were more likely to disperse, and

approximately half the surviving juveniles dispersed.

2.6 Management Implications: CWD and Long Distance Movements

2.6.1 Mule Deer

Surveillance data show that CWD positive mule deer are most often in wildlife

management zones adjacent to the S.S. River (CCWHC, unpublished data). Preliminary

analyses suggest the probability of harvesting a CWD-positive deer in Saskatchewan increases

with terrain ruggedness, proximity to major rivers and distance to roads (Rees et al. 2009). It is

generally suspected that the S. S. River basin plays a role in spread of disease westward through

Saskatchewan and into the neighboring province of Alberta. Although proximity to water was

not a significant factor in long distance movement paths when tested in the GEE, terrain

ruggedness and grassland proximity were significant and both are attributes associated with the

S. S. River. Coulees with dramatic inclines and grassland unaltered by agricultural practices exist

along the River basin and provide the rugged terrain that is a key mule deer habitat requirement

(Wood et al. 1989, Dusek 1975). Pearson’s correlation scores for terrain ruggedness with

distance to grassland and distance to water were -0.244 and -0.181 respectively, suggesting that

rugged terrain and wetlands were found not only along the river but also interspersed throughout

the landscape.

Miller et al. (2000) speculated that the CWD epidemic in Colorado and Wyoming

corresponded to patterns of mule deer migration along the North and South Platte Rivers. Long

distance movements by white-tailed deer were found to be directional according to the

orientation of a major river in Montana (Dusek et al. 1989) and South Dakota (Kernohan et al.

1994), and aligned with ridged topography (but not rivers) in Pennsylvania (Long 2006).

49

Jacques et al. (2003) suggested that CWD spread in South Dakota would be hastened by white-

tailed deer movements along river-bottom habitats.

In this study, mean migration orientation was 41°, roughly perpendicular to the

orientation of the S.S. River at the BEEMAT site. Disease spread in a northeastern direction

from CWD-positive foci alongside the River has not been systematically evaluated using

surveillance data but new cases have been found further north from the River in recent years.

This suggests migrating deer may play a role in disease spread in a northeastern direction, but

disease transmission is more efficient in mule deer habitat along the River basin itself, perhaps

due to seasonal differences in mule deer grouping behavior.

Mule deer congregate in larger groups and smaller areas during winter months (Wood et

al. 1989, Lingle 2003, Conner and Miller 2004). In the BEEMAT study area, deer moved to the

coulee habitat along the river in cold winter months and were more spatially separated in spring

and summer (Silbernagel, 2010). A few adult males were an exception—they summered along

the river, migrated north in autumn and stayed there until spring. Silbernagel (2010) also

analyzed GPS data of mule deer in this study and found that proximity between collared deer

was highest during winter months. CWD transmission risk is likely higher along the River during

winter months because deer are in higher density, and share a limited number of forage locations.

Deer that migrate from densely-populated wintering areas may carry disease to their summering

ranges.

Recent studies have shown that contact with prion-contaminated environments is an

effective mode of CWD infection (Miller et al. 2004, Johnson et al. 2006, Mathiason et al. 2009,

Wiggins 2009). Further investigation into the role of prion-contaminated soil in CWD-endemic

areas like the S. S. River basin would greatly enhance our understanding of CWD transmission.

50

2.6.2 White-tailed Deer and CWD in Saskatchewan

Although our observations of white-tailed deer movements were limited, we speculate the

risk of geographic spread of CWD is substantial for this species, and less predictable than for

mule deer in the Mixed Grassland. In 1972–74, a mark-recovery study (Stewart and Runge

1985) of white-tails in the Crystal Beach Game Preserve (CBGP) of central Saskatchewan

showed high emigration of juveniles (fawns or yearlings) and long travel distances (x̄ = 215

km). The farthest recovery was a yearling doe found in southwest Manitoba, 673 km southeast of

its natal site. Over 30% of recovered deer trapped as juveniles were located >50 km from the

study area, whereas only 1 of 19 adult-tagged deer was recovered >5 km from the study area. No

significant directionality of movement was noted. There may have been avoidance of travel

along the major highway in the area, but it was not a barrier to movement. Seven deer were

located across the South Saskatchewan River to the east and south of the study area. The results

of this study showed no barriers to white-tailed deer movement and presumable ease of disease

transmission throughout the species’ range of Saskatchewan, and beyond into neighboring

provinces and states.

Our model results suggested that mule deer movement can be predicted by variables such

as rugged terrain and grassland and it follows that habitat preferences could be similarly used to

predict white-tailed deer movement. Since white-tailed deer are better adapted to agricultural

land than are mule deer, their movement may be less limited in Saskatchewan. Chronic wasting

disease prevalence in Wisconsin white-tailed deer is related to abundance of deer habitat (Joly et

al. 2006). Wood, Mackie, and Hamlin (1989) studied sympatric mule and white-tailed

populations in Montana prairies and found compared to mule deer, white-tailed deer used grain

fields more frequently, were generally more mobile, had larger home ranges, and had less

fidelity to traditional home ranges. Our observations of white-tailed movement patterns were

51

similar. In recent years a focus of CWD has developed in white-tailed deer in north-central

Saskatchewan near Nipawin (53.37°N, -104.01°W), in the Boreal Transition ecoregion and in

proximity to the Saskatchewan River. While our study results cannot be applied to this area,

there are recent examples in the literature of white-tailed movement patterns and habitat

preferences being used to predict long-distance movements (Felix et al. 2007, Diefenbach et al.

2008, Frost et al. 2009).

52

CHAPTER 3 MULE DEER (ODOCOILEUS HEMIONUS) SURVIVAL RATES

IN A CWD-ENDEMIC AREA OF SOUTHERN SASKATCHEWAN

3.1 Abstract

Chronic wasting disease (CWD) has been detected in wild mule deer (Odocoileus

hemionus) of southern Saskatchewan since 2000 and continues to increase in geographic range

and prevalence. Management programs for the disease continue to evolve but have focused

primarily on herd reduction in known CWD areas. Deer survival rates have not been measured in

these areas and have been impacted by management decisions in recent years. Chronic wasting

disease will likely affect survival rates in the long term. Survival analyses were completed in

concurrence with a radio telemetry mule deer movement study from 2006–2009. Adult deer

mortality occurred primarily during hunting season but at differing rates between sexes

depending on the year and management program. Juveniles (up to 2 years old) had lower

survival rates than adults and they experienced higher mortality during the summer period

following capture. Radio collar mass contributed to their poor survival and in the future

researchers should use a collar that is a maximum of 2% of body mass. Deer in good body

condition were half as likely to die and deer in very poor body condition were twice as likely to

die. Home range habitat content did not appear to affect survival. Mule deer survival in

Saskatchewan was influenced primarily by harvest regulations. In the future, as CWD

prevalence increases, disease impacts on survival rates can be measured and compared to current

survival rates.

53

3.2 Introduction

Chronic wasting disease (CWD) was first detected in a wild mule deer (Odocoileus

hemionus) in Saskatchewan, Canada in 2000. The disease is contagious, fatal, difficult to

manage, and it continues to threaten wild cervids over an ever-expanding range within North

America. Long-term impacts of CWD on host populations are unclear but will depend on other

factors such as recruitment and other sources of mortality. Age and cause-specific mortality rates

of deer are required in order to predict the effect of CWD on deer populations.

Only a few recent studies have estimated the effect of CWD on deer populations. In

Colorado, where localized CWD prevalence in mule deer was 41% in adult males and 20% in

adult females, Miller et al. (2008) found that CWD-infected individuals had poorer survival than

their uninfected counterparts, and that mountain lions (Puma concolor) selectively preferred

CWD-infected mule deer prey. Prior to that publication, studies evaluating impacts of CWD on

deer populations were based on theoretic models which predicted population declines and

possible host extinction (Miller et al. 2000, Gross and Miller 2001). Models are helpful when

decisions must be made with best available knowledge, but assumptions should be rigorously

evaluated and models revised and adapted as new knowledge becomes available (Schauber and

Woolf 2003).

In North America, deer survival is affected by hunting, winter severity, disease,

predation, vehicle collisions, interspecific competition, habitat changes, population density, and

natural causes (White et al. 1987, Wood et al. 1989, Unsworth et al. 1999, DelGiudice et al.

2002). Deer experience seasonal gains and losses in body mass, depending on metabolic needs

and range conditions, and their body condition can affect their survival during strenuous winter

months (Mackie et al. 2003, Klaver et al. 2008). Our objective was to estimate mule deer

survival patterns and rates in southern Saskatchewan and to estimate the role of potential factors

54

on survival. Factors were classified as intrinsic, consisting of measures of body size and

condition, and extrinsic, consisting of characteristics of the home range which may affect

survival. The impact of radio collars on survival was also addressed. This study contributes to

our understanding of the impact of CWD on local deer population demographics, and the effects

of radio marking on deer survival.

3.3 Methods

3.3.1 Study Area

Refer to section 2.3.1 for study area description.

3.3.2 Capture

Refer to section 2.3.2 for capture methods.

3.3.3 Radio Collars

Throughout the course of the research project, we used 4 different types of radio collars.

We began with one type of VHF and one type of GPS, weighing 425 g and 900 g respectively.

After the first year we decided to compromise on battery life in order to use lighter weight

collars on juvenile deer (a different type of VHF and GPS, weighing 200 g and 300 g

respectively). Mortality seemed disproportionately high for juveniles with GPS collars in the

first year they were captured (2007). The cause was unclear but it could have been an effect of

increased predation, or added stress at a crucial time of year. We followed animal care protocol

guidelines by using collars weighing less than 5% of the deer’s body mass (Canadian Council on

Animal Care 2003), but felt that the radio collars were having a notable adverse effect on

survival in smaller-bodied deer and warranted investigation.

3.3.4 Body Condition Index (BCI)

Body condition refers to an animal’s energetic reserves in fat and muscular tissue, and is

measured by its mass relative to its structural size (Cattet 2000, Schulte-Hostedde et al. 2005).

55

Beginning in 2007, each deer’s mass was measured to the nearest kg using a cot hung from a

scale on a tripod. Body size was measured to the nearest cm for chest girth, hindfoot length

(Parker 1987), neck circumference, and total body length (Bois et al. 1997, in Lesage et al.

2001). Deer were laterally recumbent for all measures. We derived the 4 log-transformed

measures into a single body size value (PC1) using principle components analysis in SPSS

(version 17.0; SPSS Inc., Chicago, Ill.) (Schulte-Hostedde et al. 2005). Prior to calculating the

PC1 or BCI, we assessed the dataset for outliers in measurements, and removed one deer with an

obvious erroneous hindfoot length measurement and also removed 18 deer (captured in 2008)

with questionable mass measures because it was noted the scale had not been properly zeroed.

We assessed correlation of body mass and body size for linearity. If PC1 was positively

related to all body size measures, it was a good indicator of size (Dobson 1992, Schulte-

Hostedde et al. 2001, 2005). Body condition scores were the standardized residuals of the

regression of body size and body mass (Cattet et al. 2002, Schulte-Hostedde et al. 2005).

Each deer’s BCI score was used as a covariate in Cox regression. We also tested for

thresholds in body condition by assigning BCI scores to poor (25th percentile), mid, and good

(75th percentile) categories. If these thresholds were non-significant, we tested a more extreme

threshold: very poor (10th percentile) or very good (90th percentile).

A t-test modified for unequal variances was used to detect difference between 2007 and

2008 BCI.

3.3.5 Home Range Attributes

3.3.5.1 Land cover classification map

With assistance of remote sensing specialists, we developed a land cover map of the

study area using 20 m multispectral SPOT (Satellite Pour l’Observation de la Terre) imagery

(Alberta Terrestrial Imaging Centre, Lethbridge, AB, T1J 0P3). Three images taken in July 2007

56

encompassed the ANT, BEEMAT, and SWI sites, and an image taken September 2007 included

the DOU site. The latter image was classified separately because of spectral differences due to

vegetative phenology. PCI Geomatica software (PCI Geomatics, Richmond Hill, ON, L4B 1M5)

was used to orthocorrect raw imagery.

Training data were collected throughout June, July, and August 2007 on transects walked

during deer surveys. At regular intervals, observers recorded GPS locations and dominant

vegetation types within 20 m. Additional training data were acquired through visual

interpretation of the satellite imagery, aided by our knowledge of the area and confirmed with

high resolution (2.5 m) panchromatic SPOT imagery (Telus 2006). We retained 30 percent of

the training data for accuracy assessment. Overall accuracy was 90% for the DOU site and 91%

for the other sites (see appendix A).

Topographic data were stacked with the imagery to assist in classification: a 25 m digital

elevation model (Government of Canada, Centre for Topographic Information) and terrain

ruggedness raster (TRI) (Evans 2004). A homogeneity filter was applied to the near-infrared

band of the SPOT imagery to better delineate vegetative areas. Supervised classification was

performed in ENVI 4.5 software (ITT Visual Information Solutions, Boulder, CO 80301) using a

maximum likelihood classifier. Pixel values overlapped for some classes identified in the

training data and were not discernible by the software (e.g., forage crops were similar to annual

crops). These were lumped into simpler classes, resulting in a final classification scheme of 4

classes: crop, grassland, wetland (including open water) and shrub or woodland. The crop class

included fallow unvegetated fields.

3.3.5.2 Home range calculation

Home range polygons were calculated as 95% kernel density estimates (KDE) (Rodgers

et al. 2007) in ArcGIS 9.2 (Environmental Systems Research Institute, Redlands, Calif.). Least

57

squares cross validation (LSCV) was used as a smoothing factor for VHF-collared deer with

suitable sample sizes (> 10 or < 100) (Hemson et al. 2005). GPS-collared individuals had sample

sizes > 100 and for these 275 m was selected as a smoothing factor because href (software-

generated reference bandwidth) tends to over-smooth and inflate home range size (Seaman et al.

1999).

Home ranges were intersected with the land cover, digital elevation, and TRI data to

estimate habitat covariates for each deer. Mean elevation, mean terrain ruggedness, and

proportion of grass, shrub, wetland, and crop within home range were calculated.

3.3.6 Statistical Analyses

3.3.6.1 Dataset

Deer that were lost to follow-up because of collar loss, collar failure, failure to locate, or

those still collared at the end of the study period were right-censored, and those that died within

21 days of capture were excluded from analysis. Studies often exclude deaths within a short

period of capture, often 14 days but ranging from 7 to 26 (Lawrence et al. 2004, Farmer et al.

2006, Grovenburg 2007). We selected 21 based on the frequency of telemetry signal checks

post-capture. Mortality signals were investigated as soon as possible; however, cause of death

was often unknown due to scavenging. Death was confirmed by finding carcass remnants at the

collar site, including scavenged carcass, hair, bones, or blood. Collars found without any

remnants were usually fully-expanded VHF collars or were from loosely-collared juveniles and

these were considered to be slipped collars and not deaths. For each deer, the time (in months)

between capture and last observed live signal was used to calculate the research lifespan.

3.3.6.2 Kaplan-Meier estimates

Annual and seasonal survival rates were calculated following the Kaplan-Meier (K-M)

procedure (Kaplan and Meier 1958) modified for staggered entry of animals (Pollock et al.

58

1989). Survival estimates were calculated for 2006, 2007 and 2008 for adult females and for

2007 and 2008 for adult males. Survival estimates for juveniles were calculated for the first

(2007) capture cohort only, as the sample size (n = 11) of the second cohort in 2008 was low.

Annual survival was from April to March and seasons were divided into hunt (September to

December), winter (January to April) and summer (May to August). We identified a winter

period by its latter year (e.g., 2007 for the 2006–2007 winter). Differences in survival functions

between groups were tested by log-rank chi-square statistics (Pollock et al. 1989): specifically,

we used the generalized Wilcoxon test (an adaptation of the log-rank test that weights by the

number of at-risk individuals per interval) (Breslow 1970, Hosmer and Lemeshow 1999). The

log-rank test is appropriate when hazards are proportional (i.e., hazard lines do not cross) (Collett

2003). If they are not, a test developed by Breslow (1984) accounts for acceleration and is

appropriate. Z-tests are used to compare 2 survival curves at a particular time (Pollock et al.

1989). α was set at 0.05. Comparisons between sexes included only adult deer because sexes

were pooled for juvenile datasets.

3.3.6.3 Cox regression

Covariate effects on hazard rates were tested using Cox proportional hazards regression

model (Cox 1972) in PROC PHREG (version 9.2; SAS Institute, Cary, NC) (Allison 1995). The

Cox model is described by the equation:

{ }kikii xxtth ββλ ++= ...exp)()( 110 ,

Meaning the hazard (h) for individual i at time t is the product of:

(1) )(0 tλ which is the baseline hazard experienced by all animals, and

(2) the exponentiated linear set of k covariates.

59

Thus we can derive the expected hazard rate at any time (t) for any individual (i) with knowledge

of its covariates.

Survival analyses were completed at 3 levels with different sample sizes depending on

the variables tested. We used the full dataset to test significance of basic variables: age (in years,

at capture time), age class (juv or ad), capture type (helicopter or trap), collar type (GPS or

VHF), and capture year. In addition, we examined 2 subsets of the deer for different risk factors:

intrinsic and extrinsic factors.

Intrinsic variables included body condition, body mass, and ratio of radio collar mass to

body mass. We categorized body condition into percentile thresholds (Table 3.1). Three broad

categories (1st, 2nd + 3rd, and 4th quartiles) were tested first, and if there was no effect at that level

then thresholds at the 10th and 90th percentile were tested. We categorized collar-to-body-mass

ratio into less than or greater than 2% of body mass. We suspected a higher hazard risk was

experienced by small (juv) deer wearing GPS collars, and >2% appeared to fit that subgroup.

The intrinsic measures dataset excluded 2006 captures because their body measurements were

not recorded and a body condition index could not be calculated.

Table 3.1: BCI categories determined by percentile.

BCI category Percentile BCI value

very poor ≤10 ≤-1.19

poor ≤25 ≤-0.68

mid 26 to 74 -0.68 to 0.61

good ≥75 ≥0.61

very good ≥90 ≥1.34

Variables in the extrinsic factors dataset included home range parameters: migratory

strategy (migratory, resident, or dispersal), average terrain ruggedness (TRI), average elevation,

and 4 habitat type proportions: grassland, shrub, crop, and wetland. Deer with fewer than 20

locations were excluded because their home range estimates were not accurate enough and might

60

bias results.

The Cox model assumes individuals experience proportional hazards through time.

Hazard lines that cross for different strata or groups violate this assumption (Hosmer and

Lemeshow 1999, Cox 1972). Prior to regression analysis, we used PROC LIFETEST in SAS to

visually assess proportionality of survival and hazard lines and then stratified accordingly. Cox

regression analyses were performed in SAS. We tested time-dependence of significant variables

by including a term representing an interaction between a suspected covariate and time (Allison

1995). The “exact” method was used to handle ties in the dataset.

If none of the covariates explained survival better than the null model (i.e., if P > 0.1),

then the null model was accepted as the final model. We used forward stepwise regression for

inclusion of variables, and AIC values for comparison of models where ∆AIC ≤ 4.0 (Burnham

and Anderson 2002). Individually-tested variables with P values < 0.2 for their coefficients (β)

were included in multivariable analyses, and retained in final models if P < 0.10. We chose α =

0.1, as is sometimes done in regression analyses when the parameter coefficient (or hazard ratio)

is strong but not significant at the 95% confidence level (Johnson 1999, Farmer et al. 2006). If

variables were collinear, we selected one based on biological importance. We tested suspected

covariates for interactions, including body condition and collar mass. Radio collars may have

been an additional stressor for deer in poor body condition and may have exacerbated their

mortality rates. Hazard ratios (also known as risk ratios) were estimated for significant variables.

Hazard ratios indicate the hazard change per one unit change in the variable (Riggs and Pollock

1992). Most variables were dichotomous so the hazard ratio reflected risk for members vs. non-

members of a group. Model P values reported are Wald Chi-square statistics.

61

Differences in body condition between groups (e.g., sexes, capture years) were detected

by t-tests. Wellsch’s t-test was used if Levene’s test showed unequal variances. We used chi-

square to detect difference in proportions of groups assigned to BCI categories. Significance was

set at α = 0.05 for these analyses and they were performed in SPSS.

3.3.6.4 Post-hoc winter severity analysis

Winter severity was assessed post-hoc to address between-year difference in body

condition scores. Body condition was calculated based on measurements at the end of winter, so

the data from the months just prior to capture were used to describe winter severity affecting

condition at capture time. We obtained Environment Canada data from 3 weather stations located

in or near our study sites at Swift Current, Elbow, and Lucky Lake Saskatchewan (Government

of Canada 2008). We averaged their daily minimum, maximum, and mean temperatures to

represent study area weather. Snow depth data were consistently available for only the Elbow

weather station. Winter was described by data from November 1 to March 31. For each day the

minimum temperature fell below -20°C, 1 point was accumulated, and for each day snow depth

exceeded 10 cm, 1 point was accumulated (DelGiudice et al. 2002, Pitt et al. 2008). Summer

precipitation values (from May 1 to October 31) were used to represent forage growth on the

winter range (Farnes 1991). For comparison, normal values were estimated using the 30-year

average of temperature and precipitation data from 2 of the weather stations. 30-year normal

values were not available for the station at Lucky Lake, nor were 30-year snow depth data

available for any station.

3.4 Results

3.4.1 Capture

Two hundred and six individuals were included in the survival dataset after removing

white-tailed deer, confirmed capture myopathies and those collared <21 days. Of these, 89 deer

62

died and 117 were censored. Twenty-nine animals died during winter months, 48 during hunting

season months, and 11 during summer months. Subsets used for Cox regression had smaller

sample sizes: n = 140 for the body condition subset and n = 151 for home range subset.

3.4.2 CWD Results

Six individuals tested CWD-positive at time of capture. All were captured in 2007: 5

were from ANT (4 AD M, 1 AD F) and 1 was from SWI (AD M). One male died within a month

of capture and his death may have been capture-related but the carcass was too scavenged for

necropsy. The others died within 22 months of capture (range 9 to 22 months) and only one was

observed with clinical disease near its death, the others died of unknown cause. Further analysis

of the effect of CWD on survival was precluded by the low sample size.

3.4.3 Body Measures

Average body mass of adult females was 61 kg (SD = 9, n = 36); adult males 75 kg (SD =

13, n = 67); juveniles (sexes pooled) 35 kg (SD = 5, n = 80). Further details are found in Table

3.2.

Table 3.2: Body measures of captured mule deer. Juvenile sexes are pooled.

ADULT MALE ADULT FEMALE JUVENILE

x̄ SD x̄ SD x̄ SD

mass (kg) 75 13 61 9 35 5

chest circumference (cm) 102 6 98 6 79 4

body length (cm) 180 11 169 8 141 8

neck circumference (cm) 46 5 38 3 32 2

hindfoot length (cm) 51 2 48 1 43 2

n 67 36 80

63

3.4.4 Body Condition Index

PC1 scores explained 92% of the variance in body size measures. Chest girth was the

body measure most strongly correlated with body size (r = 0.97) and all body measurement

values contributed to PC1 (Table 3.3). Correlation between body size and mass was high (R2 =

0.97) (Figure 3.1). The relationship between mass and body size was:

mass (kg) = 19.84 (PC1) + 55.03

Body size and body condition were not related (r = 0.000, P = 0.99, n = 183). Scores

below -0.67 were in the poor BCI category (lowest = -2.6), and above 0.61 were in the good BCI

category (highest = 3.2) (Figure 3.2). Thresholds for very good and very poor BCI were 1.34 and

-1.19, respectively.

Table 3.3: Factor loadings of morphologic measures on PC1 from principle components analysis for mule deer body size

Body Measure Loading on eigenvector

LN_chest 0.972

LN_neck 0.951

LN_hindfoot 0.956

LN_length 0.961

64

Figure 3.1: Relationship between body size (derived from principle components analysis of 4 body measures) and mass. Juveniles and adults showed similar linearity.

65

Figure 3.2: Body condition index boxplot. Values below -0.67 and above 0.61 were in the poor and good BCI categories, respectively. Extreme values of 3 deer are also shown. 3.4.5 Radio Collars

The radio collars we used ranged from 0.4% to 3.4% of the deer’s body mass at capture

time. Most were well within the guidelines: 83% (132 of 159) were under 2% of body mass. Of

the 24 deer with a collar >2% of its mass, 14 died and 10 were censored. Eight were in poor

body condition, 5 good, and 11 mid. Nineteen were captured by helicopter and 5 by Clover trap.

All but one were captured in 2007. Nineteen were juveniles, 4 were yearlings (3 M, 1F), and 1

adult (M). There were 3 from the DOU site (2 ad, 1 juv), and 7 at each of the other sites.

66

3.4.5 Survival Estimates

3.4.5.1 Kaplan-Meier estimates

3.4.5.1.1 Adults

Annual

Adult female survival was higher in 2008 than in 2007, whereas adult male survival was

higher in 2007 than in 2008 (Table 3.4).

Table 3.4: Annual (Apr–Mar) survival rates (SE) of adult male and adult female radio-collared mule deer in southern Saskatchewan, 2006–2008.

Sex 2006 2007 2008

Female 0.76 (0.06) 0.72 (0.05) 0.86 (0.08)

Male - 0.82 (0.05) 0.62 (0.07)

Seasonal

Adult survival was lowest during the hunting season (Table 3.5), but differences in

seasonal survival between years and sexes were observed. In 2006 and 2007, adult does had high

survival during the summer, lower in the winter, and lowest in the hunt season. In hunt season

2008, female survival increased to 0.91 (SE = 0.04) and was comparable to survival of previous

winters (Figure 3.3). Male survival in 2007 was higher in the hunting season than winter season,

but their survival in 2008 was lowest during the hunting season. Male and female survival was

not proportional and was similar during summer and winter, so we tested differences between

sexes for only hunt seasons 2007 and 2008. In hunt season 2007, female survival was lower than

male (Z1 = 2.23, P < 0.025) and in 2008, male survival was lower than female (Z1 = 2.34, P <

0.01).

67

0.60

0.65

0.70

0.75

0.80

0.85

0.90

0.95

1.00

summer 2006 hunt 2006 winter 2007 summer 2007 hunt 2007 winter 2008 summer 2008 hunt 2008

Female Male

Figure 3.3: Seasonal survival rates of adult female and male mule deer with 95% confidence intervals. Table 3.5: Adult mule deer seasonal survival 2006–2008. Note the first 3 seasons were female only, otherwise males and females are pooled. SEASON SURVIVAL LOWER

95% CL

UPPER

95% CL

summer 2006 1.00 1.00 1.00

hunt 2006 0.82 0.71 0.93

winter 2007 0.90 0.84 0.96

summer 2007 0.97 0.93 1.00

hunt 2007 0.87 0.81 0.93

winter 2008 0.91 0.86 0.96

summer 2008 0.98 0.96 1.00

hunt 2008 0.83 0.76 0.90

68

3.4.5.1.2 Juveniles

Annual

Juvenile annual survival in 2007 was 0.49 (SE = 0.08) and was significantly lower than

adult survival in 2007 (0.77, SE = 0.04) (χ12 = 13.7, P < 0.001). 2008 juvenile survival rates

were poor estimates due to low sample size.

Seasonal

Juvenile survival was lower than adult survival in all 3 seasons, but was only statistically

significant in the summer of 2007 (Z1 = 2.205, P < 0.025). Survival was 0.81 (SE = 0.07) in the

summer following capture (Table 3.6). In contrast, adults experienced very little mortality during

summer periods of all 3 years.

Table 3.6: Seasonal survival of juvenile mule deer captured in 2007 SEASON SURVIVAL LOWER

95% C.L.

UPPER

95% C.L.

summer 2007 0.81 0.67 0.95

hunt 2007 0.78 0.62 0.94

winter 2008 0.86 0.73 0.99

3.4.6.1 Cox regression

Regression analyses were stratified based on sex and study site because their hazard lines

were not parallel. There were no time-dependent interactions of covariates and the variable

Capture Year was non-significant (P > 0.6); therefore, the use of right-censored data was

justified despite the staggered entry date of captured individuals.

3.4.6.1.1 Full dataset

Age class was the only variable of significance in the full dataset (χ12 = 3.17, P = 0.075).

The hazard risk of juveniles was 1.58 times higher than that of adults (95% CI = 0.96−2.60).

69

This result is comparable to the significant Wilcoxon test of K-M survival curve differences

between adults and juveniles.

3.4.6.3.1.2 Body condition subset

Deer in the good BCI (75th percentile) category had a significantly low hazard risk and

deer in the very poor BCI (10th percentile) category had a significantly high hazard risk. Deer

with collar mass >2% of body mass also had a significantly high hazard risk.

The good body condition (BCI Good) category was most explanatory of survival (P <

0.06, β = -0.62, SE = 0.33). Deer in good body condition were at a lower hazard risk (0.54, 95%

CI = 0.28–1.02). Deer with collars >2% of their body mass were at a higher risk of death (1.82,

95% CI = 0.96–3.46). The best model included both variables (Table 3.7) and made the hazard

ratios slightly more conservative than when individually tested (Table 3.8). This was likely

caused by deer that were within both a good or poor category of body condition and had collar

mass >2%. Twenty-four deer had collars >2% of body mass: 8 in poor BCI, 5 in good BCI, and

11 in mid BCI. All hazard ratios overlapped 1 and should be viewed with caution.

Body condition index did not interact with collar mass (P > 0.42), nor was there

interaction between collar >2% of body mass and poor BCI (P > 0.77) or body condition index

and collar >2% (P > 0.59).

Mean body condition of deer captured in 2007 (0.88, SE = 0.9, n = 120) differed (t138 =

2.25, P < 0.03) from deer captured in 2008 (-0.47, SE = 0.2, n = 20).

70

Table 3.7: Model results of Cox regression analysis

Model P χ2 df -2 Log L AIC Δ AIC wi

Null - - 0 306.30 306.30 2.84 0.07

Collar >2% 0.07 3.39 1 303.19 305.19 1.73 0.12

BCI very poor 0.06 3.53 1 303.09 305.09 1.63 0.13

BCI very poor, Collar >2% 0.04 6.41 2 300.50 304.50 1.03 0.18

BCI good 0.06 3.66 1 302.20 304.20 0.73 0.20

BCI good, Collar >2% 0.04 6.66 2 299.46 303.46 0 0.29

Table 3.8: Cox regression model parameter estimates and hazard ratios with individual parameters and combined.

Model Label Parameter

Hazard

Ratio

95% Hazard Ratio

Confidence Limits β

collar >2% 1.75 0.93 3.32 0.56 1

collar and

BCI good good BCI 0.55 0.29 1.04 -0.60

2 BCI good good BCI 0.54 0.28 1.02 -0.62

lower 10% BCI 1.90 0.91 3.98 0.64 3

collar and

BCI very poor collar >2% 1.73 0.91 3.27 0.55

4 BCI very poor lower 10% BCI 2.01 0.97 4.18 0.70

5 collar collar >2% 1.82 0.96 3.46 0.60

3.4.6.3.1.3 Home range subset

The null model was accepted as the best compared to models with home range covariate

measures. Significance values for variables ranged from P = 0.23 (average elevation) to P = 0.88

(proportion of wetland). Most P values were above 0.5.

3.4.6.3.1.4 Winter severity

Winter of 2008 was more severe than that of 2007 according to temperature, snow depth,

and rainfall measures (Table 3.9). These results help explain why deer captured in 2008 were in

poorer body condition than those captured in 2007. Thirty-year normal data are listed where

available. Summer rainfall was below normal in both years.

71

Table 3.9: Comparison of winter severity indicators for 2007 and 2008 and 30-year normals. Capture

year

Winter below

-20°C

(days)

Snow depth

>10 cm

(days)

summer

rainfall

(mm)

2007 2006/07 30 75 238

2008 2007/08 38 93 208

- 30-yr. normal 34 n/a 352

3.5 Discussion

3.5.1 CWD Management Program

Most of the radio-collared deer died during hunting season months. Hunters were notified

of the CWD research project in the annually-published Saskatchewan Hunters’ & Trappers’

Guide, and were asked not to shoot collared deer, but it was not illegal to do so. Twenty-four

deer were shot by hunters who reported and returned the collars. We suspect some of the other

deer that died during the hunting season may have been fatally wounded or not recovered by

hunters. Hunters often reported not seeing a collar at the time of shooting and others indicated

they had not read the related information in the guide. Some hunters reported seeing a deer with

a collar and as a result not shooting. I suspect the number of radio-collared deer that were shot

under-represents the hunting loss in the area, but the effect is probably marginal.

3.5.1.1 Male and female survival during the hunting season

The provincial CWD management program changed dramatically beginning in 2008.

The CWD area expanded to include more management zones but permits were less liberal. The

Earn-a-Buck Program was altered such that antlerless mule deer permits were no longer provided

for free but at a cost of $19.62 each and the fee for an either-sex mule deer license was increased

from $19.81 to $37.29. Heads from antlerless deer, submitted in order to receive an either-sex

permit, were now accepted from anywhere in the province whereas previously they were

required to be taken from CWD zones only. In 2007, hunters could take an unlimited number of

72

antlerless deer but in 2008 they were limited to 4 antlerless deer. Consequently, adult doe

survivorship increased in 2008 hunt season.

The explanation for decreased survival of males in 2008 compared to 2007 is less clear.

Three individuals died in autumn season 2007, and 15 in 2008; of these, 1 died by vehicle

collision, 8 were hunter-shot, one was suspected to have been shot, and the cause of death for the

remaining 8 were unknown as only scavenged remains were found. It cannot be assumed that

autumn mortality is entirely due to hunting; however, it is likely several of the deaths of

unknown-cause were a result of wounding during the hunting season. It is possible that hunters

were more likely to submit the antlerless heads now that they could be harvested province-wide

and more either-sex licenses were purchased as a result. The antlerless permits in 2007 included

a 3-point-or-less deer, but not in 2008. This may have resulted in fewer sub-adult males in the

population in 2008 and added pressure on mature males. Adult male survival during the 2007

hunt season was high considering this was a herd reduction area (0.94, SE = 0.03). Conclusions

drawn from this result are limited, because only 2 years of data were available. There could be

random factors at play, and factors we did not or could not measure (e.g., age structure of the

herd, hunter attitudes, effect of radio collar on likelihood of harvest).

Annual survival rates of adult male mule deer averaged 0.72 (SD = 0.14), which is

relatively high compared to other studies. In Montana, adult male survival rates varied from 1990

to 1995 but averaged 0.52 (SD = 0.13) and 0.57 (SD = 0.09) on 2 study areas, the latter having

some private land with restricted access during hunting season (Pac and White 2007). In Idaho,

adult male annual survival rates over a 4-year period starting in 1993 averaged 0.54 (SD = 0.15)

(Bishop et al. 2005). Even though our study areas were within herd reduction zones with liberal

deer licenses, hunting pressure in Saskatchewan remains limited by a low hunter population

73

relative to other regions. To date, herd reduction for CWD management has relied on incentives

to increase hunting in affected areas; given these findings, alternative methods such as the use of

sharp-shooters, etc. may be required to achieve desired population goals.

3.5.2 Body Condition

Body condition scores measure nutritional status of ungulates, and are good predictors of

survival. Our results indicated that deer in good body condition (75th percentile, n = 37) were

more likely to survive, and deer in very poor body condition (10th percentile, n = 14) were more

likely to die. In this study, the threshold of body condition that reduces survival probability was

more extreme than that which increases survival probability. In a longer study we might find that

the effect of body condition on survival varies depending on other factors, such as winter

severity. The lower threshold of poor body condition that affects survival might increase if

circumstances were more strenuous.

Compared to 30-year normals, winters during the study period had similar numbers of

days below -20°C, but snow depth data were unavailable for comparison. Summer precipitation

values appeared to be below normal. However, this conclusion is based on data that may not

accurately represent the study area. The central weather station at Lucky Lake did not have 30-

year normal precipitation data available and the 2 other stations were quite varied from one

another: 254 mm and 450 mm were the values for the Swift Current and Elbow weather stations

respectively. Elbow falls in the moist mixed grassland ecoregion, which receives more rainfall

than the mixed grassland ecoregion that encompasses most of the study area. These statistics

might give some evidence of below normal precipitation that could affect body condition, but the

evidence is not strong enough to make a valid scientific assessment.

Body condition varied between years and corresponded with measures of winter severity.

Years of extreme snow depth have been related to white-tailed deer mortality from malnutrition

74

in Saskatchewan (Runge and Wobeser 1975), but areas where mule deer range generally have

less snow accumulation due to wind and terrain.

Males were generally in poorer condition than females, which we expected because male

mule deer expend their energy reserves during the breeding season and are more likely to

succumb to winter starvation than their female or non-breeding male counterparts (Geist 1994a).

3.5.3 Radio Collars

Guideline 28 of the Canadian Council on Animal Care (CCAC)’s care and use of wildlife

(2003) states that radio transmitter weight should not exceed 5% of the animal’s body mass. Our

results indicate survival effects can be seen at levels below 5% and suggest that the limit should

be lowered to ~2%. Collar weight may not be the only risk factor as deer with collars between 2

and 3.4% of body mass were mostly juveniles captured in 2007 and other unmeasured factors

may have contributed to the high hazard rate. The capture and collaring protocol differed

between years, leaving no control to compare the treatment in 2007 and 2008. Of the deer with

collars >2%, all but 1 were captured in 2007, and the one captured in 2008 was an adult. There

were few adults with collars >2% of body mass. However, we encourage researchers to further

investigate the effect of collars on survival and behavior and to limit collar mass to below 2% of

the research animal’s body mass whenever possible.

3.5.4 Home Range Habitat Effect on Survival

None of the habitat measures we evaluated affected survival. However, habitat selection

occurs at multiple scales and is influenced by numerous factors (e.g., predator abundance,

conspecific use of habitats, forage quality) (Klaver et al. 2008). A more specific resource-

selection approach may have been more appropriate, one that measures frequency of use

compared to availability (Garshelis 2000). Another approach was undertaken by Bender et al.

(2007b), who related habitat selection to mule deer doe condition in New Mexico. Mortality risk

75

by predator species varies by habitat, especially in areas where there are large predators such as

mountain lions or wolves (Bishop et al. 2005, McLoughlin et al. 2005), but can also be affected

by conspecific group size. Coyotes are the largest common predator in southern Saskatchewan

and it is more likely they would take down fawns than yearlings or adults. In our study areas,

some lands are poorly accessible to hunters due to rough terrain or restricted access and may give

protection to deer during hunting season.

3.6 Management Implications

Survival rates depict radio-collared deer in Southern Saskatchewan for a 2- or 3-year

period. There were management practices in effect for the CWD program that are not in place

throughout the entire province; therefore, these rates are specific only to these areas and cannot

describe province-wide mule deer survival rates. We have demonstrated survival differences

between adult males and females dependent on harvest regulations. In the future, managers may

want to consider how survival rates changed as a result of program adjustments between 2007

and 2008 prior to making new adjustments to the program. Juvenile deer had poor survival

compared to adults. The status of the population growth (or decline) cannot be inferred without

fawn survival estimates.

Hazard ratios that approximate 1 indicate no effect of a covariate. Ninety-five per cent

confidence intervals were near 1 at their limits for most significant parameters, and results

should be treated with some caution. Hazard ratios could be more accurate if the study was

repeated and/or sample sizes increased.

Wildlife survival depends on a number of factors, none of which are static. Weather

factors such as winter severity play an important role in overwinter survival in deer, and a 3-year

window of observation is not likely to accurately estimate the range in winter conditions or

drought conditions Saskatchewan endures. These survival estimates are a snapshot in time and

76

should be treated as such. They will be useful in the future to compare with survival rates if

CWD prevalence rates increase to a point it is affecting herd health. We recommend ongoing

study of deer survival in CWD-endemic areas of Saskatchewan, with particular emphasis on the

knowledge gaps of fawn survival rates and effects of CWD on survival.

77

CHAPTER 4 SYNTHESIS

Wildlife managers and research biologists in North America have been struggling to

contain CWD since it was first found in a wild cervids in the 1980s (Williams and Miller 2002).

Despite their efforts, no management agency has effectively eradicated the disease and it

continues to spread geographically and increase in prevalence. At the time of its emergence in

wild populations, there was little known about CWD etiology. The main course of action in

North America has been to reduce deer densities where CWD was known to exist and to sample

adjacent areas to detect new cases. As we gain knowledge about prion resistance to degradation

and its persistence in the environment (Mathiason et al. 2009), the complexity of managing

CWD in wild cervids further unfolds. Because eradication is improbable, prevention of disease is

the best method to protect wild cervids.

Researchers need to identify and quantify transmission risk from multiple causes so that

managers can use best available knowledge to prevent CWD in areas where it doesn’t yet exist.

Risks include those associated with farmed cervids, environmental contamination, and wild deer

ecology. Area-specific knowledge of host movement, contact rates and types, survival, and

habitat use are vital to understanding, predicting, and modeling disease transmission and spread.

In Saskatchewan, CWD was first found in mule deer and prevalence in mule deer remains higher

than in white-tailed deer. Our research objective was to evaluate long distance movements and

survival rates of radio-collared mule deer along the South Saskatchewan River basin.

78

We have concluded that mule deer of any age-sex class are capable of travelling long

distances and that their movements may be predicted by habitat preferences. Fifty-five per cent

of juveniles emigrated a median distance of 22.8 km (SE = 13.6) and most dispersal events

occurred in spring. Forty-two per cent of adults migrated a median distance of 16.0 km (SE =

2.6) and 45% of these were obligate migrants. Males and females migrated at a similar rate and

similar distances, but obligate migrants were more likely to be female. In the BEEMAT study

site, most (68%) of the adult deer were migrants and at the other study sites 24 to 30% were

migrants. Temporary excursions averaged 9.0 km (SE = 0.8) and were frequent year-round by

females and most often during autumn by males. Deer travelled long distances in a short time

period: travel was completed an average 1.1 days per migration and 5.5 days per dispersal.

Migration movements were significantly directional in the BEEMAT study area, and dispersal

movements were not random but were not significantly directional. Long distance movement

paths were associated with selection of habitats with rugged terrain and in close proximity to

grassland.

Based on these results, managers should consider that deer in expansive grassland

habitats are more likely to be migrants than those in fragmented habitats, and their movement

orientation will follow rugged terrain and grassland habitat. In this area, the average mule deer

home range size is 24.1 km2 (SE = 1.7) (Silbernagel 2010), but the overall area that may be

included in a deer’s movement patterns is larger when considering seasonal, temporary, or

dispersal movements. Disease transmission and spread may occur over a large scale along with

deer movements. Seasonality of movements and behaviors should also be considered because

deer are more likely to come into contact with one another during the winter months (Conner et

79

al. 2008, Silbernagel 2010) and deer in this study area tended to winter in the coulees of the

South Saskatchewan River.

Autumn (hunting season) mortality is currently most predictive of mule deer survival in

Saskatchewan. Juvenile survival was lower than adult. Relative to other areas, survival rates of

mule deer in Saskatchewan were relatively high. Because harvest rates can be adjusted,

managers should be able to maintain desired mule deer population goals even if mortality rates

increase in areas of high CWD prevalence. Deer in good body condition had lower mortality

rates than their counterparts in average or below average body condition. Deer with radio collars

that weighed more than 2% of the animal’s body weight had lower survival rates than those with

lighter collars. Researchers should use the smallest collars available that meet their needs. With

this knowledge of current survival trends, it will be possible in the future to assess the impact of

chronic wasting disease on mule deer survival rates.

4.1 Study Limitations

We have identified mule deer migration and dispersal patterns in southern Saskatchewan.

However, white-tailed deer are the more common species throughout much of the province, and

transmission risks associated with their behavioral ecology should be assessed and considered.

Researchers from the University of Alberta conducted a white-tailed deer research project

concurrent with ours that will produce information on contacts between groups and other risks

associated with CWD transmission. White-tailed deer movement patterns have been described in

a game preserve in central Saskatchewan (Stewart and Runge 1985) and of adult does in an

agricultural area of southeastern Saskatchewan (Brewster and Longmuir 1994). Literature and

research from the United States can also be used to identify and assess risks associated with

white-tailed deer.

80

Conclusions about juvenile survival and dispersal rates were informative but would have

been improved with a bigger sample size and with more years of data. Because most of the deer

were captured in 2007, there may have been factors associated with that year (e.g., weather,

harvest rates) that influenced the outcome, especially of survival rates. Survival analyses with

multiple years of data provide better information than short-term studies.

Body condition indices are often validated with comparison to body content from

carcasses. This allows more detailed analysis of the type of body energy being stored (e.g., body

water, muscle, fat) (Cattet et al. 2002, Stephenson et al. 2002). For the purposes of this study this

was not necessary, but would have been more informative and useful for comparison with

similar methods in the future or for comparison with mule deer from different regions.

The effect of increased hunting pressure in herd reduction areas for CWD management

on mule deer movement behavior was not addressed in this study. Hunting season begins in

September and peaks in November, and coincides with rutting behavior. We could not separate

the effects of each on mule deer movements but in the future, a study design could include areas

of varied hunting pressure for comparison of movement behaviors. Kufeld et al. (1988) found

that female mule deer did not move outside their home ranges in response to hunting pressure,

but did seek areas of greater cover.

4.2 Conclusion

Chronic wasting disease is ecologically complex and wildlife managers struggle to

understand and balance the various ecologic, social, and economic factors associated with the

disease and its management. To date, attempts to reduce transmission and geographic spread of

CWD have been unsuccessful. This research has identified factors which may help explain past

failures and provide guidelines for future management actions. This research begins to derive

population survival estimates in CWD-endemic areas in order to predict long-term effects of

81

CWD on mule deer populations. In addition, this information can be used to evaluate effects and

transmission of other diseases or pathogens in wild deer.

More specifically, we found that mule deer migration or dispersal movements were more

frequent and extensive than previously thought. Mule deer often travel 16 to 23 km (and not

infrequently up to 200 km) and are more likely to be migratory in areas with extensive grassland.

They are likely to follow rugged terrain and grassland, which may help to explain the apparent

westerly spread of CWD along the South Saskatchewan River basin into Alberta and would

suggest further spread along river corridors and north within the Missouri Coteau. Obviously

movement patterns of the host are important in understanding the spread of CWD; however,

other factors need to be evaluated in order to understand the ecology of this complex disease.

These factors include differential survival of dispersing animals, rates of infections in dispersers,

spread of CWD by predators and scavengers, infection rates and movement of sympatric CWD-

susceptible cervids, and environmental persistence, to name a few.

Survival analysis estimates in CWD management areas indicated that management

programs to increase hunting pressure did not reduce survival rates below values reported

elsewhere from non-herd reduction areas. Alternative methods of herd reduction will likely be

needed in order to reach targets.

We found that the current guideline of <5% for radio collar mass to body mass for mule

deer to be too high and that radio collars that weighed >2% of mule deer body mass negatively

affected survival and recommend that researchers use transmitters below this threshold. GPS

collars, relative to VHF transmitters, have larger and heavier batteries in order to obtain more

precise and frequent data. Larger batteries also allow for longer tracking periods before collars

need to be replaced. Researchers should carefully consider the tradeoff between data quality and

82

quantity and the negative effect of collar mass on the study animal’s survival and welfare. As

technology improves, collar mass will likely decrease and so will negative effects on the animal,

but in the short-term, large heavy batteries/collars should be avoided.

83

Literature cited

Acton D. F., G. A. Padbury, and C. T. Stushnoff. 1998. The Ecoregions of Saskatchewan. Canadian Plains Research Center, University of Regina, Regina, Saskatchewan, Canada.

Alexander K. A., J. W. McNutt. 2010. Human behavior influences infectious disease emergence at the human-animal interface. Frontiers in Ecology and the Environment eView.

Allison P. D. 1995. Survival analysis using the SAS system: a practical guide. SAS publishing, Cary, North Carolina, USA.

Anderson A. E., E. Medin, and D. C. Bowden. 1972. Indices of carcass fat in a Colorado mule deer population. The Journal of Wildlife Management 36:579–594.

Augustine D. J., S. J. McNaughton. 1998. Ungulate effects on the functional species composition of plant communities: herbivore selectivity and plant tolerance. The Journal of Wildlife Management 62:1165–1183.

Bandy P. J., I. M. Cowan, and A. J. Wood. 1970. Comparative growth in four races of black-tailed deer (Odocoileus hemionus). Part I. Growth in body weight. Canadian Journal of Zoology 48:1401–1410.

Bender L. C., L. A. Lomas, and J. Browning. 2007a. Condition, survival, and cause-specific mortality of adult female mule deer in North-Central New Mexico. Journal of Wildlife Management 71:1118–1124.

Bender L. C., L. A. Lomas, and T. Kamienski. 2007b. Habitat effects on condition of doe mule deer in arid mixed woodland-grassland. Rangeland Ecology & Management 60:277–284.

Bishop C. J., J. W. Unsworth, and E. O. Garton. 2005. Mule deer survival among adjacent populations in southwest Idaho. Journal of Wildlife Management 69:311–321.

Bois C., M. Crete, J. Huot, and J. P. Ouellet. 1997. Predicting body composition of white-tailed deer (Odocoileus virginianus) carcasses using body mass and morphological measurements. Canadian Journal of Zoology/Revue Canadien de Zoologie 75:1790–1795.

Bollinger T., P. Caley, E. Merrill, F. Messier, M. W. Miller, M. D. Samuel, and E. Vanopdenbosch. 2004. Chronic wasting disease in Canadian wildlife: An expert opinion on the epidemiology and risks to wild deer. Final report on a meeting of the expert scientific panel on chronic wasting disease, July 2004. University of Saskatchewan, Saskatoon, Saskatchewan, Canada.

Breslow N. E. 1970. A generalized Kruskal-Wallis test for comparing K samples subject to unequal patterns of censorship. Biometrika 57:579.

Breslow N. E., L. Edler, and J. Berger. 1984. A two-sample censored-data rank test for acceleration. Biometrics 40:1049–1062.

84

Brewster D.A., Longmuir J.R. 1994. Movement patterns and habitat preferences of white-tailed deer in Saskatchewan. Saskatchewan Environment and Resource Management, Regina, SK, Canada.

Brinkman T. J., C. S. Deperno, J. A. Jenks, B. S. Haroldson, and R. G. Osborn. 2005. Movement of female white-tailed deer: Effects of climate and intensive row-crop agriculture. Journal of Wildlife Management 69:1099–1111.

Brown C. G. 1992. Movement and migration patterns of mule deer in southeastern Idaho. The Journal of Wildlife Management 56:246–253.

Bruggeman J. E., R. A. Garrott, P. J. White, F. G. R. Watson, and R. Wallen. 2007. Covariates affecting spatial variability in bison travel behavior in Yellowstone National Park. Ecological Applications 17:1411–1423.

Bunnell F. L., A. S. Harestad. 1983. Dispersal and dispersion of black-tailed deer: models and observations. Journal of mammalogy 64:201–209.

Burnham K. P., D. R. Anderson. 2002. Model selection and multimodel inference: a practical information-theoretic approach. Springer Verlag, New York, New York, USA.

Canadian Council on Animal Care. 2003. Guidelines on: the care and use of wildlife. Ottawa, Ontario, Canada. Available online at http://www.ccac.ca/en/CCAC_Programs/Guidelines_Policies/GDLINES/Wildlife/Wildlife.pdf

Canadian Plains Research Center. 2005. The Encyclopedia of Saskatchewan. University of Regina, Regina, Saskatchewan, Canada.

Cattet M. 2000. Biochemical and physiological aspects of obesity, high fat diet, and prolonged fasting in free-ranging polar bears. Ph.D. Thesis, University of Saskatchewan, Saskatoon, Saskatchewan, Canada.

Cattet M. R. L., N. A. Caulkett, M. E. Obbard, and G. B. Stenhouse. 2002. A body-condition index for ursids. Canadian Journal of Zoology/Revue Canadienne de Zoologie 80:1156–1161.

Clover M. R. 1956. Single-gate deer trap. California Fish and Game 42:199–201.

Collett D. 2003. Modelling survival data in medical research. CRC press, Boca Raton, Florida, USA.

Conner M. M., M. W. Miller. 2004. Movement patterns and spatial epidemiology of a prion disease in mule deer population units. Ecological Applications 14:1870–1881.

Conner M. M., M. R. Ebinger, J. A. Blanchong, and P. C. Cross. 2008. Infectious disease in cervids of North America: data, models, and management challenges. Annals of the New York Academy of Sciences 1134:146–172.

85

Conner M. M., M. W. Miller, M. R. Ebinger, and K. P. Burnham. 2007. A meta-BACI approach for evaluating management intervention on chronic wasting disease in mule deer. Ecological Applications 17:140–153.

Conover M. R. 1997. Monetary and intangible valuation of deer in the United States. Wildlife Society Bulletin 25:298–305.

Côté S. D., M. Festa-Bianchet, and F. Fournier. 1998. Life-history effects of chemical immobilization and radiocollars on mountain goats. The Journal of Wildlife Management 62:745-752.

Côté S. D., T. P. Rooney, J. P. Tremblay, C. Dussault, and D. M. Waller. 2004. Ecological impacts of deer overabundance. Annual review of Ecology, Evolution, and Systematics 35:113–147.

Cox D. R. 1972. Regression models and life-tables. Journal of the Royal Statistical Society.Series B (Methodological) 34:187–220.

Cypher B. L. 1997. Effects of radiocollars on San Joaquin kit foxes. The Journal of Wildlife Management 61:1412-1423.

DelGiudice G. D., M. R. Riggs, P. Joly, and W. Pan. 2002. Winter severity, survival, and cause-specific mortality of female white-tailed deer in north-central Minnesota. The Journal of Wildlife Management 66:698–717.

Derek Murray Consulting Associates. 2006. Economic evaluation of hunting in Saskatchewan. Final report prepared for Saskatchewan Environment, December, 2006.Regina, Saskatchewan, Canada.

Diefenbach D. R., E. S. Long, C. S. Rosenberry, B. D. Wallingford, and D. R. Smith. 2008. Modeling distribution of dispersal distances in male white-tailed deer. Journal of Wildlife Management 72:1296–1303.

Dobson F. S. 1982. Competition for mates and predominant juvenile male dispersal in mammals. Animal Behaviour 30:1183–1192.

Dusek G. L. 1975. Range relations of mule deer and cattle in prairie habitat. The Journal of Wildlife Management 39:605–616.

Dusek G. L., R. J. MacKie, J. D. Herriges Jr, and B. B. Compton. 1989. Population ecology of white-tailed deer along the lower Yellowstone River. Wildlife Monographs 104:3–68.

Evans J. 2004. Topographic ruggedness index script. http://arcscripts.esri.com/details.asp?dbid = 12435

Farmer C. J., D. K. Person, and R. T. Bowyer. 2006. Risk factors and mortality of black-tailed deer in a managed forest landscape. The Journal of Wildlife Management 70:1403–1415.

Farnes, P.E. A scaled index of winter severity. 59th Western snow conference proceedings. April 12–15, 1991. Juneau, Alaska, USA.

86

Felix A. B., D. P. Walsh, B. D. Hughey, H. Campa, and S. R. Winterstein. 2007. Applying landscape-scale habitat-potential models to understand deer spatial structure and movement patterns. Journal of Wildlife Management 71:804–810.

Frost C. J., S. E. Hygnstrom, A. J. Tyre, K. M. Eskridge, D. M. Baasch, J. R. Boner, G. M. Clements, J. M. Gilsdorf, T. C. Kinsell, and K. C. Vercauteren. 2009. Probabilistic movement model with emigration simulates movements of deer in Nebraska, 1990–2006. Ecological Modelling 220:2481–2490.

Garrott R. A., R. M. Bartmann, and G. C. White. 1985. Comparison of radio-transmitter packages relative to deer fawn mortality. The Journal of Wildlife Management 49:758–759.

Garrott R. A., G. C. White, R. M. Bartmann, L. H. Carpenter, and A. W. Alldredge. 1987. Movements of female mule deer in northwest Colorado. The Journal of Wildlife Management 51:634–643.

Garshelis D. L. 2000. Delusions in habitat evaluation: measuring use, selection, and importance. Research techniques in animal ecology: controversies and consequences p.111–164.

Garson G. D. 2009. Generalized Linear Models and Generalized Estimating Equations. from Statnotes: Topics in Multivariate Analysis. Updated 6/26/2009. Accessed 2/22/2010. http://faculty.chass.ncsu.edu/garson/PA765/gzlm_gee.htm

Geist V. 1994a. Mule and black-tailed deer: sparring, courtship, and mating. Pages 252–265 In

D. Gerlach, S. Atwater, and J. Schnell, editors. Deer, Stackpole Books, Mechanicsburg, PA, USA.

Geist V. 1994b. Mule and black-tailed deer: social behavior. Pages 278–285 In D. Gerlach, S. Atwater, and J. Schnell, editors. Deer, Stackpole Books, Mechanicsburg, PA, USA.

Government of Canada. 2008. National climate data and information archive. Updated 11/1/2008. Accessed 4/30/2010. http://www.climate.weatheroffice.gc.ca/climate_normals/index_e.html

Government of Saskatchewan. 2009. Crop Report. Updated 12/30/2009. Accessed 1/4/2010. http://www.agriculture.gov.sk.ca/Crop-Report

Grear D. A., M. D. Samuel, J. A. Langenberg, and D. Keane. 2006. Demographic patterns and harvest vulnerability of chronic wasting disease infected white-tailed deer in Wisconsin. Journal of Wildlife Management 70:546–553.

Gross J. E., M. W. Miller. 2001. Chronic wasting disease in mule deer: Disease dynamics and control. Journal of Wildlife Management 65:205–215.

Grovenburg T.W. 2007. Movement patterns of white-tailed deer in east central South Dakota relative to winter ranges and management unit boundaries. M.Sc. Thesis, South Dakota State University. Brookings, South Dakota, USA.

Happ G. M., H. J. Huson, K. B. Beckmen, and L. J. Kennedy. 2007. Prion protein genes in caribou from Alaska. Journal of wildlife diseases 43:224–228.

87

Hawkins R. E., W. D. Klimstra. 1970. A preliminary study of the social organization of white-tailed deer. The Journal of Wildlife Management 34:407–419.

Hedlund J. D. 1975. Tagging mule deer fawns in south-central Washington, 1969–1974. Northwest Science 49:153–157.

Hemson G., P. Johnson, A. South, R. Kenward, R. Ripley, and D. MacDonald. 2005. Are kernels the mustard? Data from global positioning system (GPS) collars suggests problems for kernel home-range analyses with least-squares cross-validation. Journal of Animal Ecology 74:455–463.

Holzenbein S., R. L. Marchinton. 1992. Emigration and mortality in orphaned male white-tailed deer. The Journal of Wildlife Management 56:147–153.

Horwitz P., B. A. Wilcox. 2005. Parasites, ecosystems and sustainability: an ecological and complex systems perspective. International Journal for Parasitology 35:725–732.

Hosmer D. W., S. Lemeshow. 1999. Applied survival analysis: regression modeling of time to event data. John Wiley and Sons, New York, New York, USA.

Jacques C. N., J. A. Jenks, A. L. Jenny, and S. L. Griffin. 2003. Prevalence of chronic wasting disease and bovine tuberculosis in free-ranging deer and elk in South Dakota. Journal of Wildlife Diseases 39:29–34.

Jenness J. 2005. Alternate animal movement routes extension for ArcView 3.2, version 2.1. Jenness Enterprises. http://www.jennessent.com/arcview/alternate_routes.htm

Johnson C. J., K. E. Phillips, P. T. Schramm, D. McKenzie, J. M. Aiken, and J. A. Pedersen. 2006. Prions adhere to soil minerals and remain infectious. PLoS Pathog 2:e32.

Johnson D. H. 1999. The insignificance of statistical significance testing. The Journal of Wildlife Management 63:763–772.

Joly D. O., M. D. Samuel, J. A. Langenberg, J. A. Blanchong, C. A. Batha, R. E. Rolley, D. P. Keane, and C. A. Ribic. 2006. Spatial epidemiology of chronic wasting disease in Wisconsin white-tailed deer. Journal of Wildlife Diseases 42:578–588.

Kammermeyer K. E., R. L. Marchinton. 1976. Notes on dispersal of male white-tailed deer. Journal of Mammalogy 57:776–778.

Kaplan E. L., P. Meier. 1958. Nonparametric estimation from incomplete observations. Journal of the American Statistical Association 53:457–481.

Kenward R. E., S. S. Walls, and K. H. Hodder. 2001. Life path analysis: scaling indicates priming effects of social and habitat factors on dispersal distances. Journal of Animal Ecology 70:1–13.

Kernohan B. J., J. A. Jenks, and D. E. Naugle. 1994. Movement patterns of white-tailed deer at Sand Lake National Wildlife Refuge, South Dakota. Prairie Naturalist 26:293–293.

88

Kie J. G., B. Czech. 2000. Mule and black-tailed deer. Pages 629–657 In S. Demarais and P. Krausman, editors. Ecology and management of large mammals in North America. Prentice Hall, Upper Saddle River, New Jersey, USA.

Klaver R. W., J. A. Jenks, C. S. Deperno, and S. L. Griffin. 2008. Associating seasonal range characteristics with survival of female white-tailed deer. Journal of Wildlife Management 72:343–353.

Kovach W. 2009. Oriana version 3.0. Kovach Computing Services, Anglesey, Wales, United Kingdom.

Krausman P. R., J. J. Hervert, and L. L. Ordway. 1985. Capturing deer and mountain sheep with a net-gun. Wildlife Society Bulletin 13:71–73.

Kucera T. E. 1978. Social behavior and breeding system of the desert mule deer. Journal of Mammalogy 59:463–476.

Kufeld R. C., D. C. Bowden, and D. L. Schrupp. 1988. Influence of hunting on movements of female mule deer. Journal of Range Management 41:70–72.

Kufeld R. C., D. C. Bowden, and D. L. Schrupp. 1989. Distribution and movements of female mule deer in the Rocky Mountain foothills. The Journal of Wildlife Management 53:871–877.

Lawrence R. K., S. Demarais, R. A. Relyea, S. P. Haskell, W. B. Ballard, and T. L. Clark. 2004. Desert mule deer survival in southwest Texas. Journal of Wildlife Management 68:561–569.

Lesage L., M. Crête, J. Huot, and J. P. Ouellet. 2001. Evidence for a trade-off between growth and body reserves in northern white-tailed deer. Oecologia 126:30–41.

Lingle S. 2003. Group composition and cohesion in sympatric white-tailed deer and mule deer. Canadian Journal of Zoology 81:1119–1130.

Long E. S. 2006. Landscape and demographic influences on dispersal of white-tailed deer. Ph.D. thesis, Pennsylvania State University, University Park, USA.

Long E. S., D. R. Diefenbach, C. S. Rosenberry, and B. D. Wallingford. 2008. Multiple proximate and ultimate causes of natal dispersal in white-tailed deer. Behavioral Ecology 19:1235–1242.

Long E. S., D. R. Diefenbach, C. S. Rosenberry, B. D. Wallingford, and M. D. Grund. 2005. Forest cover influences dispersal distance of white-tailed deer. Journal of Mammalogy 86:623–629.

Mackie R. J., J. G. Kie, D. F. Pac, and K. L. Hamlin. 2003. Mule deer. Pages 889–905 In G. A. Feldhamer, B. A. Thompson, and J. A. Chapman, editors. Wild mammals of North America: biology, management, and conservation, John Hopkins University Press, Baltimore, Maryland, USA.

89

Mathiason C. K., S. A. Hays, J. Powers, J. Hayes-Klug, J. Langenberg, S. J. Dahmes, D. A. Osborn, K. V. Miller, R. J. Warren, and G. L. Mason. 2009. Infectious prions in pre-clinical deer and transmission of chronic wasting disease solely by environmental exposure. PloS one 4:e5916.

Mathiason C. K., J. G. Powers, S. J. Dahmes, D. A. Osborn, K. V. Miller, R. J. Warren, G. L. Mason, S. A. Hays, J. Hayes-Klug, D. M. Seelig, M. A. Wild, L. L. Wolfe, T. R. Spraker, M. W. Miller, C. J. Sigurdson, G. C. Telling, and E. A. Hoover. 2006. Infectious prions in the saliva and blood of deer with chronic wasting disease. Science 314:133–136.

McLoughlin P. D., J. S. Dunford, and S. Boutin. 2005. Relating predation mortality to broad-scale habitat selection. Journal of Animal Ecology 74:701–707.

Mech L. D. 1983. Handbook of animal radio-tracking. University of Minnesota Press, Minneapolis, Minnesota, USA.

Miller C. A., L. B. Shelby. 2009. Hunters' general disease risk sensitivity and behaviors associated with chronic wasting disease. Human Dimensions of Wildlife 14:133–141.

Miller M. W., E. S. Williams, N. T. Hobbs, and L. L. Wolfe. 2004. Environmental sources of prion transmission in mule deer. Emerging Infectious Diseases 10:1003–1006.

Miller M. W., H. M. Swanson, L. L. Wolfe, F. G. Quartarone, S. L. Huwer, C. H. Southwick, and P. M. Lukacs. 2008. Lions and prions and deer demise. PLoS ONE 3:e4019.

Miller M. W., E. S. Williams, C. W. McCarty, T. R. Spraker, T. J. Kreeger, C. T. Larsen, and E. T. Thorne. 2000. Epizootiology of chronic wasting disease in free-ranging cervids in Colorado and Wyoming. Journal of Wildlife Diseases 36:676–690.

Needham M. D., J. J. Vaske. 2008. Hunter perceptions of similarity and trust in wildlife agencies and personal risk associated with chronic wasting disease. Society & Natural Resources 21:197–214.

Nelson M. E., L. D. Mech. 1984. Home-range formation and dispersal of deer in northeastern Minnesota. Journal of Mammalogy 65:567–575.

Nelson M. E. 1995.Winter range arrival and departure of white-tailed deer in northeastern Minnesota. Canadian Journal of Zoology 73:1069 - 1076

Nicholson M. C., R. T. Bowyer, and J. G. Kie. 1997. Habitat selection and survival of mule deer: tradeoffs associated with migration. Journal of Mammalogy 78:483–504.

Nixon C. M., L. P. Hansesn, P. A. Brewer, J. E. Chelsvig, J. B. Sullivan, T. L. Esker, R. Koerkenmeier, D. R. Etter, J. Cline, and J. A. Thomas. 1994. Behavior, dispersal, and survival of male white-tailed deer in Illinois. Illinois Natural History Survey Biological Notes 139:1–29.

Nixon C. M., P. C. Mankin, D. R. Etter, L. P. Hansen, P. A. Brewer, J. E. Chelsvig, T. L. Esker, and J. B. Sullivan. 2007. White-tailed deer dispersal behavior in an agricultural environment. The American Midland Naturalist 157:212–220.

90

Noss R. F., B. Csuti. 1997. Habitat fragmentation. Pages 269–304 In G. K. Meffe and C. R. Carroll, editors. Principles of Conservation Biology. Sinauer Associates, Inc., Sunderland, Massachusetts, USA.

Osnas E. E., D. M. Heisey, R. E. Rolley, and M. D. Samuel. 2009. Spatial and temporal patterns of chronic wasting disease: fine-scale mapping of a wildlife epidemic in Wisconsin. Ecological Applications 19:1311–1322.

Ozaga J. J., L. J. Verme, and C. S. Bienz. 1982. Parturition behavior and territoriality in white-tailed deer: impact on neonatal mortality. Journal of Wildlife Management 46:1–11.

Pac D. F., G. C. White. 2007. Survival and cause-specific mortality of male mule deer under different hunting regulations in the Bridger Mountains, Montana. Journal of Wildlife Management 71:816–827.

Pan W. 2001. Akaike's information criterion in generalized estimating equations. Biometrics 57:120–125.

Parker K. L. 1987. Body-surface measurements of mule deer and elk. The Journal of Wildlife Management 51:630–633.

Pitt J. A., S. Larivière, and F. Messier. 2008. Survival and body condition of raccoons at the edge of the range. Journal of Wildlife Management 72:389–395.

Pollock K. H., S. R. Winterstein, C. M. Bunck, and P. D. Curtis. 1989. Survival analysis in telemetry studies: the staggered entry design. Journal of Wildlife Management 53:7–15.

Porter W. F., N. E. Mathews, H. B. Underwood, R. W. Sage, and D. F. Behrend. 1991. Social organization in deer: implications for localized management. Environmental management 15:809–814.

Rees E. E., E. H. Merrill, T. K. Bollinger, Y. T. Hwang, M. J. Pybus, and D. W. Coltman. 2009. Modeling CWD spread using surveillance data. 3rd International Chronic Wasting Disease Symposium p.54.

Riggs M. R., K. H. Pollock. 1992. A risk ratio approach to multivariable analysis of survival in longitudinal studies of wildlife populations. Pages 74–89 In D.R. McCullough and R.H. Barrett, editors. Wildlife 2001: populations. Elsevier Applied Science, New York, NY, USA.

Robinette W. L. 1966. Mule deer home range and dispersal in Utah. The Journal of Wildlife Management 30:335–349.

Robinette W. L., D. A. Jones, G. Rogers, and J. S. Gashwiler. 1957. Notes on tooth development and wear for Rocky Mountain mule deer. The Journal of Wildlife Management 21:134–153.

Rodgers A. R., A. P. Carr, H. L. Beyer, L. Smith, and J. G. Kie. 2007. HRT: Home Range Tools for ArcGIS. Version 1.1.

Rosenberry C. S., M. C. Conner, and R. A. Lancia. 2001. Behavior and dispersal of white-tailed deer during the breeding season. Canadian Journal of Zoology 79:171–174.

91

Runge W., G. Wobeser. 1975. A survey of deer winter mortality in Saskatchewan. Report for Saskatchewan Department of Tourism and Renewable Resources, Regina, Saskatchewan, Canada.

Russell G. S., D. J. Levitin. 1995. An expanded table of probability values for Rao's Spacing Test. Communications in Statistics-Simulation and Computation 24:879–888.

Sabine D. L., S. F. Morrison, H. A. Whitlaw, W. B. Ballard, G. J. Forbes, and J. Bowman. 2002. Migration behavior of white-tailed deer under varying winter climate regimes in New Brunswick. The Journal of Wildlife Management 66:718–728.

Saskatchewan Ministry of Environment. 2010. Current distribution maps of CWD positive wild deer in Saskatchewan. http://www.environment.gov.sk.ca/cwd_distribution

Sawyer H., F. Lindzey, and D. McWhirter. 2005. Mule deer and pronghorn migration in western Wyoming. Wildlife Society Bulletin 33:1266–1273.

Schauber E. M., A. Woolf. 2003. Chronic wasting disease in deer and elk: a critique of current models and their application. Wildlife Society Bulletin 31:610–616.

Schuler K. L., J. A. Jenks, C. S. DePerno, M. A. Wild, and C. C. Swanson. 2005. Tonsillar biopsy test for chronic wasting disease: two sampling approaches in mule deer and white-tailed deer. Journal of Wildlife Diseases 41:820–824.

Schulte-Hostedde A. I., B. Zinner, J. S. Millar, and G. J. Hickling. 2005. Restitution of mass-size residuals: validating body condition indices. Ecology 86:155–163.

Seaman D. E., J. J. Millspaugh, B. J. Kernohan, G. C. Brundige, K. J. Raedeke, and R. A. Gitzen. 1999. Effects of sample size on kernel home range estimates. The Journal of Wildlife Management 63:739–747.

Severinghaus C. W. 1949. Tooth development and wear as criteria of age in white-tailed deer. The Journal of Wildlife Management 13:195–216.

Shelton P., N. E. Mateus-Pinilla, H. Weng, P. McDonald, M. O. Ruiz, W. Brown, J. Novakofski, T. Beissel, M. Douglas, and M. Douglas. 2009. Sharpshooting implementation and efficacy in management of CWD in Illinois. 3rd International chronic wasting disease symposium p. 49.

Silbernagel E.R. 2010. Factors affecting movement patterns of mule deer (Odocoileus hemionus) in southern Saskatchewan: implications for chronic wasting disease spread. M.Sc. Thesis, University of Saskatchewan, Saskatoon, SK, Canada.

Skuldt L. H., N. E. Mathews, and A. M. Oyer. 2008. White-tailed deer movements in a chronic wasting disease area in South-Central Wisconsin. Journal of Wildlife Management 72:1156–1160.

Stephenson T. R., V. C. Bleich, B. M. Pierce, and G. P. Mulcahy. 2002. Validation of mule deer body composition using in vivo and post-mortem indices of nutritional condition. Wildlife Society Bulletin 30:557–564.

92

Stewart R. R., W. Runge. 1985. The white-tailed deer of the Crystal Beach game preserve: an ecological investigation. Saskatchewan Parks and Renewable Resources, Regina, SK, Canada.

Unsworth J. W., D. F. Pac, G. C. White, and R. M. Bartmann. 1999. Mule deer survival in Colorado, Idaho, and Montana. The Journal of Wildlife Management 63:315–326.

Vercauteren K. C., S. E. Hygnstrom. 1998. Effects of agricultural activities and hunting on home ranges of female white-tailed deer. The Journal of Wildlife Management 62:280–285.

White G. C., R. A. Garrott. 1990. Analysis of wildlife radio-tracking data. Academic Press, New York, New York, USA.

White G. C., R. A. Garrott, R. M. Bartmann, L. H. Carpenter, and A. W. Alldredge. 1987. Survival of mule deer in northwest Colorado. Journal of Wildlife Management 51:852–859.

Wiens J. A. 2001. The landscape context of dispersal. Pages 96–109 In J. Clobert, E. Danchin, A. A. Dhondt, and J. D. Nichols, editors. Dispersal, Oxford University Press, Oxford, New York.

Wiggins R. C. 2009. Prion stability and infectivity in the environment. Neurochemical Research 34:158–168.

Williams E. S., M. W. Miller. 2002. Chronic wasting disease in deer and elk in North America. Revue scientifique et technique (International Office of Epizootics) 21:305–316.

Winterstein S. R., K. H. Pollock, and C. M. Bunck. 2001. Analysis of survival data from radiotelemetry studies. Pages 351–380 In J. J. Millspaugh and J. M. Marzluff, editors. Radio tracking and animal populations, Academic Press, San Diego, CA, USA.

Wolff J. O. 1993. What is the role of adults in mammalian juvenile dispersal? Oikos 68:173–176.

Wood A. K., R. J. Mackie, and K. L. Hamlin. 1989. Ecology of sympatric populations of mule deer and white-tailed deer in a Prairie environment. Wildlife Division, Montana Dept. of Fish, Wildlife & Parks, Bozeman, Montana.

Zar J. H. 1999. Biostatistical analysis. Simon & Shuster, Upper Saddle River, New Jersey.

93

Appendix

A. Accuracy Assessment of Land Cover Map

Error matrices for classified land cover maps

All data are displayed in number of pixels unless otherwise noted

Sites: ANT, BEEMAT, SWI Training data

Crop Grassland Shrub/

woodland Wetland total

3Commission

error

4User

accuracy

(%)

Crop 1349 1 0 2 1352 3 99.7

Grassland 0 1676 20 5 1701 25 98.5

Shrub/woodland 11 27 30 44 112 82 26.8

Wetland 0 16 4 22 42 20 52.4

Map

Total 1360 1720 54 73 3207 130 91.3

1Omission error 11 44 24 51 130

2Producer

accuracy (%) 99 97 56 30 91

Site:DOU Training data

Crop Grassland Shrub/

woodland Wetland total Commission

error

User

accuracy

(%)

Crop 1707 2 0 0 1709 2 99.9

Grassland 31 119 3 0 153 34 77.8 Shrub/woodland 42 89 284 34 449 165 63.3

Wetland 457 48 8 4820 5333 513 90.4

Map

Total 2237 258 295 4854 7644 714 90.4

Omission error 530 139 11 34 714

Producer

accuracy (%) 76 46 96 99 90

1Errors of ommission depict pixels that were incorrectly classed compared to known pixels of that class

2Producer accuracy: percentage of reference (training data) pixels classified correctly

3Errors of commission depict pixels with known values but were classed incorrectly

4User accuracy: percentage of classified pixels in the map that agree with reference data